← the graph
Sector 11 · 51 sub-themes

Enterprise Adoption & Products

Anthropic's anticipated IPO at a $2 trillion-plus valuation — potentially the largest stock market debut in history — is the headline signal of how dramatically the enterprise AI market has repriced. Its annualized revenue is projected to hit $100–120B by year-end, and the Ramp AI Index confirms its enterprise penetration is already ahead of OpenAI at 43.5% vs. 39.7% of U.S. businesses. OpenAI is not standing still: its own run rate doubled to $40B ARR, it deployed GPT-5.6 through a new IBM partnership targeting finance, government, and telecom, and it's recruiting heavily — including Google DeepMind's Noam Shazeer — ahead of its own IPO. Meanwhile, Google DeepMind is mid-crisis: Demis Hassabis stepped down as CEO, top Gemini leads departed to co-found a startup, and AlphaFold Nobel co-winner John Jumper left for Anthropic. Yet Google's Gemini app hit 1 billion monthly active users — faster than any prior Google product — with ChatGPT's web traffic share falling 23 points over 12 months as competition fragments the market. The mid-tier model race is intensifying: Gemini 3.7 Flash, Claude Sonnet 5, and GPT-5.6 Terra are all competing directly at similar capability and price points. DeepSeek's move — launching V4-Pro while raising API prices up to 12x — signals that the "race to zero" pricing narrative has real limits even for Chinese rivals. On the product side, Lovable's $400M raise at $13.3B and CodeRabbit's $143M at $1.5B together define a maturing stack: AI generates code at scale, and a verification layer is now a standalone billion-dollar business.

AnthropicOpenAIGoogle DeepMindGeminiClaude Sonnet 5GPT-5.6Noam ShazeerDemis HassabisJohn JumperDeepSeekLovableCodeRabbitenterprise AI market valuationAI company IPO racelarge language model competitionAI code generation toolsmid-tier model pricing warenterprise AI adoption benchmarksAI talent exodusChinese AI rivals pricingAI drug discoverygenerative AI product ecosystem
11.1

Anthropic investors target $2 trillion-plus valuation for October IPO

  • Anthropic investors are reportedly targeting a $2 trillion-plus valuation for an IPO as early as October, which would surpass SpaceX's record valuation of approximately $1.77 trillion and make it the largest stock market debut in history, according to the Financial Times and Wall Street Journal. [2][5][7]
  • The Wall Street Journal reported the target IPO window as late September to early October, with investors anticipating a concrete near-term public offering. [3][4]
  • Analysts cited in reporting expect Anthropic's annualized revenue to reach $100–120 billion by end of 2026; at a 30x valuation multiple, that would put the company's market cap as high as $3 trillion. [6]
  • Anthropic is currently valued at close to $1 trillion ahead of the anticipated IPO, reflecting its rapid growth in the AI sector. [1]
AnthropicSpaceXFinancial TimesWall Street JournalAnthropic IPOAI company valuationstock market debutrevenue projectionspublic offering timelineAI sector growthinvestor expectations
Sources
  1. [1]CNBC — Anthropic'slongstanding push for regulation is coming back to bite it. Theartificial intelligencecompany, now valued at close to $1 trillion★news
  2. [2]unusual_whales — BREAKING: Anthropic investors are betting the company could reach a $2 trillion-plus valuation in an October IPO, potentially making it the ★twitter
  3. [3]unusual_whales — BREAKING: Anthropic is looking to IPO in September or early October, per WSJ★twitter
  4. [4]unusual_whales — BREAKING: Anthropic is looking to IPO in September or early October, per WSJ★twitter
  5. [5]coinbureau — JUST NOW: Anthropic could overtake SpaceX as the world’s largest IPO at a $2T+ valuation, investors told FT. Investors say booming AI demandtwitter
  6. [6]BlockcastNews — AI新創公司Anthropic投資人據報目標其首次公開募股（IPO）估值能達2兆美元，最快10月上市。此估值將超越SpaceX，成為史上最大IPO案。分析師預期，Anthropic年化收入至2026年底有望達1000至1200億美元，若以30倍估值計算，公司市值上看3兆美元。twitter
  7. [7]AGTPinsights — Anthropic's investors are reportedly eyeing a $2 trillion+ valuation for an October IPO, which would make it the largest stock market debut twitter
11.2

Google Gemini Surpasses 1 Billion Monthly Active Users

  • Google CEO Sundar Pichai announced on August 11 that the Gemini app has surpassed 1 billion monthly active users, describing it as Google's fastest-growing product ever and the 14th Google product to reach the 1-billion milestone. The figure refers specifically to the Gemini app's MAU, separate from Gemini integrations within Google Search's AI Mode, Workspace, or YouTube. [4][5][7]
  • Google also disclosed that its Search AI Mode had separately surpassed 1 billion MAU (announced in May), while competitor ChatGPT crossed 1 billion MAU in June, according to Reuters — placing both products at comparable scale. [5][7]
  • Google released detailed Gemini usage metrics alongside the milestone: 63% of users interact with Gemini via voice, with "voice-only" interaction growing; 1 in 5 Gemini Live sessions involves sharing live camera footage or screen content; and over 150 million images are generated daily on the platform. [6]
  • On iOS, Gemini has more than 100 million active users, and macOS power users submit prompts at twice the frequency of users on other platforms; Android users can connect Gemini to 40+ apps for automated actions such as booking rides and reservations. [6]
  • GM announced that Google Gemini AI will be integrated into Cadillac and Corvette vehicles, extending Gemini's enterprise and automotive product footprint. [3]
  • Google's Gemini Spark agent received updates including a macOS launch and expanded connected apps support, broadening the product's cross-platform reach. [1]
  • Gemini Flash now includes a native thinking capability that previously required a separate Gemini 2.5 model, available directly through the Gemini API. [2]
GoogleSundar PichaiGeminiChatGPTGMCadillacCorvetteGemini SparkGemini FlashGemini APIGemini app monthly active users milestoneAI assistant growth and adoptionvoice interaction and multimodal usageautomotive AI integrationcross-platform AI expansionGemini vs ChatGPT scale comparisonAI image generation volumeconnected apps and automation
Sources
  1. [1]blog.google — Gemini Spark updates: macOS launch, connected apps and more★news
  2. [2]Ana Maria Constantin — The capability that previously required a separate Gemini 2.5 model is now a native tool inside Flash, available through the Gemini API and★news
  3. [3]GM Authority — GM Announces Google Gemini AI For Cadillac And Corvette★news
  4. [4]sundarpichai — 1B+ people are now using @Geminiapp every month to spark new ideas and get things done. It’s our fastest growing product ever, and our 14th ★twitter
  5. [5]itmedia_news — GeminiアプリのMAUが10億人を突破 Googleで14番目の大台到達製品に https://t.co/GfznA6bp1ltwitter
  6. [6]ktai_watch — Google、「Gemini」の月間アクティブユーザーが10億人を突破 音声利用が63％など https://t.co/623cKyX1F2 https://t.co/uZlyt2ykFNtwitter
  7. [7]itmedia_news — GeminiアプリのMAUが10億人を突破 Googleで14番目の大台到達製品に https://t.co/GfznA6bp1ltwitter
11.3

Anthropic in talks to acquire Israeli AI startup Decart for $6 billion

  • Bloomberg and Reuters reported that Anthropic is in advanced talks to acquire Decart AI, an Israeli AI infrastructure startup backed by Nvidia, for approximately $6 billion — which would be Anthropic's largest acquisition to date. Negotiations are described as at an early stage, meaning the deal could still change or fall through. [3][5][4]
  • Decart develops AI infrastructure and optimization technologies, and its products include Lucy (a model for real-time video editing) and Oasis (which creates simulated environments for training and testing robotics and autonomous-driving systems). [4]
  • If completed, the acquisition would give Anthropic access to Decart's technology and engineering expertise to improve the speed and efficiency of its AI systems and make better use of available computing capacity; Decart's team is expected to join Anthropic's performance and inference division. [4]
  • Decart announced it had raised $300 million in a funding round led by Radical Ventures, with Nvidia participating as a new investor, underscoring growing interest in the startup's AI infrastructure technology. [4]
  • Artefact and Anthropic announced a partnership aimed at accelerating enterprise AI adoption, combining Artefact's consulting capabilities with Anthropic's Claude models to help businesses deploy AI solutions. [1][2]
AnthropicDecart AINvidiaBloombergReutersLucyOasisRadical VenturesArtefactClaudeAI infrastructure acquisitionenterprise AI adoptioninference optimizationrobotics simulationreal-time video editingAI startup fundingcomputing efficiencyconsulting partnership
Sources
  1. [1]AiThority — Artefact and Anthropic team up to accelerate Enterprise AI Adoption★news
  2. [2]EIN News — Artefact and Anthropic team up to accelerate Enterprise AI Adoption★news
  3. [3]AviKaner — Anthropic is in talks to acquire Israeli AI company Decart for $6 billion. It would be Anthropic’s largest acquisition and would establish atwitter
  4. [4]i24NEWS_EN — 🔴 Anthropic in negotiations to acquire Israeli startup Decart AI for $6 billion - report Read more here 👉 https://t.co/AahE7vrK3Utwitter
  5. [5]shiringhaffary — NEW: Anthropic is in talks to acquire world-model and inference optimization Israel-based AI startup Decart for $6 billion, per sources w/ @twitter
11.4

Morgan Stanley sets $300 target, $600 bull case for SpaceX citing Cursor

  • Morgan Stanley analyst Adam Jonas maintained a $300 base price target on SpaceX ($SPCX) with a $600 bull case, citing Cursor as "a major factor bridging" the upside, following SpaceX's planned ~$60B acquisition of the AI coding tool. [2][3][4][5]
  • Morgan Stanley's note indicated that SpaceX's Cursor deal could open the door to a broader AI opportunity, with speculation that an xAI/Cursor combination could reach $10B ARR and a $1T valuation. [1][5]
Morgan StanleyAdam JonasSpaceXCursorxAISpaceX acquisitionAI coding toolsanalyst price targetARR valuationxAI speculationtech deal analysisbull case scenario
Sources
  1. [1]Brian Wang — Home»Artificial intelligence»XAI and Cursor Strategy to Catch Up and Beat Anthropic in Coding Cursor is training its own Opus-class model fr★news
  2. [2]DeItaone — $SPCX - SPACEX: MORGAN STANLEY SEES $300 TARGET, $600 BULL CASE Morgan Stanley maintains its $300 SpaceX target, with a $600 bull case, folltwitter
  3. [3]DeItaone — $SPCX - SPACEX: MORGAN STANLEY SEES $300 TARGET, $600 BULL CASE Morgan Stanley maintains its $300 SpaceX target, with a $600 bull case, folltwitter
  4. [4]StockSavvyShay — Morgan Stanley sees a path to a $600 bull case for $SPCX saying Cursor could be “a major factor bridging” the upside as SpaceX builds a broatwitter
  5. [5]herbertong — 🚨 MORGAN STANLEY SEES $600 BULL CASE FOR $SPCX Morgan Stanley says @SpaceX’s $60B Cursor deal could open the door to a much bigger AI opporttwitter
11.5

DeepSeek launches V4-Pro and sharply raises API prices

  • DeepSeek announced the official production release of DeepSeek-V4-Pro on August 13, simultaneously with a sweeping API price overhaul effective August 17, framing the price changes as tied to the new model launch and surging infrastructure costs driven by demand growth. [3][5]
  • The new pricing for DeepSeek-V4-Pro sets standard (off-peak) rates at $0.022/1M tokens input (cache hit), $0.66/1M tokens input (cache miss), and $1.98/1M tokens output — representing increases of approximately 12.1x (cache hit input), 3.0x (cache miss input), and 4.5x (output) versus prior rates. [3][4][6]
  • For DeepSeek-V4-Flash, the new standard rates are $0.007/1M tokens input (cache hit), $0.22/1M tokens input (cache miss), and $0.66/1M tokens output, with V4-Flash output pricing during peak hours rising to $1.32/1M tokens. [1][3]
  • DeepSeek is simultaneously introducing a peak/off-peak pricing structure: peak hours carry rates double the standard price, and are applied during two daily windows (10:00–13:00 and 15:00–19:00 Japan time), meaning the maximum per-unit price increase reaches up to 12x compared to previous pricing at peak times. [3][5]
  • The V4-Pro model carries specific agent-oriented upgrades, with the production version designated DeepSeek-V4-Pro-0813 in the source. [2]
DeepSeekDeepSeek-V4-ProDeepSeek-V4-FlashDeepSeek-V4-Pro-0813DeepSeek V4-Pro releaseAPI pricing overhaultoken pricing increasespeak/off-peak pricing structurecache hit vs cache miss pricinginfrastructure cost pressuresagent-oriented model upgradesLLM API pricing tiers
Sources
  1. [1]wallstengine — DeepSeek is sharply raising prices for its V4 models starting Aug. 16. V4-Flash output pricing during peak hours rises to $1.32 per 1M tokentwitter
  2. [2]WesRoth — DeepSeek just officially launched DeepSeek-V4-Pro — with major upgrades aimed squarely at AI agents. The production model, DeepSeek-V4-Pro-0twitter
  3. [3]itm_aiplus — 中国DeepSeek、API料金を「最大12倍」に値上げ ピーク時価格も導入 8月17日から https://t.co/ofERFuaMLDtwitter
  4. [4]itm_aiplus — 中国DeepSeek、API料金を「最大12倍」に値上げ ピーク時価格も導入 8月17日から https://t.co/ofERFuaMLDtwitter
  5. [5]nikkei — 中国DeepSeek、利用料最大12倍値上げ 需要拡大でインフラ費用転嫁 https://t.co/6PiWZDt0eetwitter
  6. [6]itm_aiplus — 中国DeepSeek、API料金を「最大12倍」に値上げ ピーク時価格も導入 8月17日から https://t.co/ofERFuaMLDtwitter
11.6

Microsoft merges consumer and business Copilot apps, retires five features

  • Microsoft is merging its previously separate consumer "Microsoft Copilot" app and the workplace-oriented "Microsoft 365 Copilot" app into a single unified application, with a refreshed UI that allows sign-in from both personal and work/school accounts; the consolidated app also enables direct access to Word, Excel, Outlook, email, calendar, and cloud storage from within the interface, while Microsoft 365 subscribers receive access to advanced AI models and higher usage limits than free users. [6][7][9][10]
  • Five features are being retired by August 18, 2026 as part of the consolidation: Group Chat (threads, messages, and media generated within group chats will not carry forward and have no named replacement), AI-generated podcasts (existing episodes can only be preserved via manual download before the cutoff), consumer-tier Deep Research (Microsoft 365 Premium subscribers are directed to "Researcher" as a replacement), Copilot Labs experimental features, and the animated Mico character. [8][9][11][12]
  • Microsoft launched a new AI services division called "Frontier" staffed with approximately 6,000 employees and backed by a $2.5 billion investment, aimed specifically at accelerating enterprise AI adoption and helping businesses deploy AI at scale. [1][2][4][5]
  • Microsoft published details on a Dynamics 365 Commerce MCP (Model Context Protocol) Server designed to enable AI agents to interact with commerce data and workflows, reflecting the company's push to embed agentic AI capabilities into its enterprise software stack. [3]
MicrosoftMicrosoft CopilotMicrosoft 365 CopilotFrontierDynamics 365 CommerceModel Context ProtocolMicoMicrosoft 365AI assistant consolidationenterprise AI adoptionagentic AI integrationconsumer and workplace app mergerfeature deprecationModel Context Protocol serverMicrosoft Frontier divisioncommerce workflow automation
Sources
  1. [1]Indiatimes — Microsoft launches new 6,000-employee AI division to boost business AI deployment★news
  2. [2]citybiz — Microsoft Launches AI Services Unit With $2.5 Billion Investment to Accelerate Enterprise Adoption★news
  3. [3]Microsoft — Dynamics 365 Commerce MCP Server for AI Agents and Agentic Commerce★news
  4. [4]entARABI — Microsoft Launches Frontier Company with $2.5 Billion Investment to Drive Enterprise AI Adoption★news
  5. [5]domain-b.com — Microsoft launches Frontier to help enterprises accelerate AI adoption★news
  6. [6]pc_watch — Microsoft、個人・ビジネスCopilotアプリを統合。Deep Researchなど廃止 https://t.co/kqeBUZGMcY https://t.co/i7sNzL52HWtwitter
  7. [7]pc_watch — Microsoft、個人・ビジネスCopilotアプリを統合。Deep Researchなど廃止 https://t.co/kqeBUZGMcY https://t.co/i7sNzL52HWtwitter
  8. [8]tech_beat_news — Microsoft will merge its consumer and Microsoft 365 Copilot apps, retiring Mico, Group Chats, podcasts, Labs and Deep Research on August 18,twitter
  9. [9]NeworldAI — Microsoft is cutting the AI bloat. It’s merging its consumer and business Copilot apps while killing features like Group Chats, AI podcasts,twitter
  10. [10]Techmitra2912 — Microsoft is merging Copilot and Microsoft 365 Copilot into one app. 🧵 Rollout starts mid-August, and 5 features are being cut — including Ptwitter
  11. [11]64bitnerd — Microsoft is merging its consumer and business Copilot apps while dropping Group Chats, AI podcasts, Deep Research and the Mico character. Ttwitter
  12. [12]genztechblog — Microsoft cuts Mico, podcasts, Group Chat, Labs and consumer Deep Research from Copilot on August 18, merging its two apps into one. That cotwitter
11.7

CodeRabbit raises $143M Series C at $1.5B valuation for AI code review

  • CodeRabbit, an AI-powered code review platform, raised $143 million in a Series C round at a $1.5 billion valuation, with Nvidia participating as a notable backer. The company processes over 2 million code reviews per week across more than 15,000 customers. [1][2][3][4][5]
  • CodeRabbit's funding thesis centers on the argument that AI-accelerated code generation has created a downstream bottleneck in code review and quality assurance — positioning the company as the judgment and trust layer for AI-generated code rather than a generation tool itself. [6][7][8][10]
  • Observers noted the raise alongside Lovable's $400 million raise at a $13.3 billion valuation as a paired signal: one company (Lovable) addresses making software generation faster via AI, while CodeRabbit addresses who reviews and ensures quality of that AI-generated output. [9]
CodeRabbitNvidiaLovableAI code reviewautomated code quality assuranceAI-generated code oversightdeveloper tools fundingSeries C investmentcode generation bottlenecksoftware quality layerAI coding infrastructure
Sources
  1. [1]business — CodeRabbit, an artificial intelligence-powered code review service, has raised $143 million in a new funding round, tapping into demand for ★twitter
  2. [2]harjotsgill — We've raised a $143M Series C at a $1.5B valuation. In 2023 we spent our time arguing that AI code needed its own reviewer. That argument is★twitter
  3. [3]aakashgupta — CodeRabbit runs 2M+ code reviews a week across 15,000+ customers. Today that vantage turned into a $143M Series C at a $1.5B valuation, and twitter
  4. [4]AGTPinsights — CodeRabbit just raised $143 million. Here's what you need to know. The AI code review startup announced a Series C funding round on August 1twitter
  5. [5]EconomicTimes — AI code review platform CodeRabbit valued at $1.5 billion in latest funding round https://t.co/4WcvulI0P2twitter
  6. [6]WesRoth — CodeRabbit just raised $143M at a $1.5 BILLION valuation and it’s betting that AI coding is creating an entirely new bottleneck: JUDGMENT. Ttwitter
  7. [7]KushalVijay_ — @coderabbitai just announced $143M Series C funding round at a $1.5B valuation. The speed at which AI code is being generated, review is thetwitter
  8. [8]LearnWithBishal — $143M raised. But the most interesting part isn’t the funding. It’s the shift they’re betting on: AI made code abundant, now trust and changtwitter
  9. [9]leoleeai — 这两天有两条 AI 软件融资，值得放在一起看： 1、Lovable 融资 4 亿美元，估值 133 亿美元。解决的是：让更多人用 AI 更快生成软件。 2、CodeRabbit 融资 1.43 亿美元，估值 15 亿美元。解决的是：AI 生成的代码越来越多以后，谁来做代码审查和质twitter
  10. [10]Itz_Steavean — In an era of endless AI code generation, the bottleneck moved downstream to PR reviews, and safety. Relying on CodeRabbit consistently in prtwitter
11.8

OpenAI Revenue Doubles to $40B Run Rate Ahead of IPO

  • OpenAI's annualized revenue run rate surpassed $40 billion (approximately $3 billion per month), doubling from its end-of-2025 level of roughly $20 billion ARR, driven by strong enterprise AI sales, according to people familiar with the matter cited by Nikkei. [1][2][6]
  • The Nikkei article specifically attributes the revenue growth to expanded corporate/enterprise AI product sales, with OpenAI chasing rival Anthropic — which reportedly hit a $40 billion run rate earlier — as it targets a potential IPO expected around 2027. [6][2]
  • OpenAI conducted a $7 billion employee share buyback at an $852 billion valuation, and has reportedly filed confidentially for an IPO, according to Kalshi and CoinUpOfficials. [4][5]
  • Polymarket reported the $7 billion transaction as a share sale/buyback of employee shares, framing it as a step ahead of a potential IPO. [3]
OpenAIAnthropicNikkeiKalshiPolymarketCoinUpOfficialsOpenAI revenue growthannualized revenue run rateenterprise AI salesIPO preparationemployee share buybackvaluationAnthropic competitioncorporate AI adoption
Sources
  1. [1]IPONewsroom_ — OPENAI'S REVENUE RUN-RATE HAS TOPPED $40 BILLION AHEAD OF ITS IPO That is roughly double where it ended 2025, when the CFO put the figure ab★twitter
  2. [2]edzitron — so OpenAI hit $40 billion in run rate (so like $3bn in monthly revenue) months after Anthropic did, mysteriously dropping the same day that twitter
  3. [3]Polymarket — JUST IN: OpenAI buys back ~$7,000,000,000.00 in employee shares ahead of a potential IPO.twitter
  4. [4]Kalshi — JUST IN: OpenAI closes $7 billion share sale at $852 billion valuation 22% chance it announces IPO this yeartwitter
  5. [5]CoinUpOfficials — One of the most anticipated IPOs of the AI era could be coming. OpenAI has reportedly filed confidentially for an IPO. Its latest valuation twitter
  6. [6]SOU_BTC — OpenAIの年換算売上高、25年末から倍増 法人向け好調で6.3兆円に - 日本経済新聞 米オープンAIの年換算売上高が400億ドル（約6兆3700億円）を超えたことが13日わかった。法人向けの人工知能（AI）販売を伸ばし、2025年末時点から倍増した。 https://t.ctwitter
11.9

OpenAI and Anthropic slash AI model prices amid Chinese rival pressure

  • OpenAI and Anthropic have both cut prices on their AI models in response to Chinese AI competitors — including DeepSeek and Moonshot — gaining ground in the market. [2][3][4][5]
  • Steve Eisman, the trader whose contrarian bets against the housing market were chronicled in "The Big Short," warned that the AI trade's Achilles' heel is over-reliance on OpenAI and Anthropic, estimating that roughly 70% of all AI revenue from Microsoft, Amazon, Google, and Oracle flows from just those two companies. [1]
  • Eisman specifically flagged a price war with cheaper Chinese models as a key risk scenario: if Chinese rivals take market share and force Anthropic and OpenAI into destructive pricing competition, he warned "we have a problem." [1]
  • Researcher and independent journalist Ed Zitron separately warned that OpenAI could become the "Lehman Brothers of the AI trade" — a systemic risk capable of taking down other companies — a concern Eisman's analysis echoes. [1]
OpenAIAnthropicDeepSeekMoonshotSteve EismanMicrosoftAmazonGoogleOracleEd ZitronLehman BrothersAI model price warChinese AI competitorsmarket concentration riskAI revenue dependencyOpenAI systemic riskcompetitive pricing pressureAI trade investment riskmarket share disruption
Sources
  1. [1]BusinessInsider — OpenAI and Anthropic getting into a price war with rivals is a threat to the AI trade, Eisman said. https://t.co/1JvYC6LFfLtwitter
  2. [2]FT — OpenAI and Anthropic in price war as Chinese AI rivals gain ground https://t.co/f3eiQkn49stwitter
  3. [3]Cointelegraph — ⚡️ LATEST: OpenAI and Anthropic slash prices on their AI models as Chinese rivals like DeepSeek and Moonshot pressure the market. https://t.twitter
  4. [4]FT — OpenAI and Anthropic in price war as Chinese AI rivals gain ground https://t.co/f3eiQkn49stwitter
  5. [5]FT — OpenAI and Anthropic in price war as Chinese AI rivals gain ground https://t.co/f3eiQkn49stwitter
11.10

Ramp Index Shows Anthropic Enterprise Gains But Fable 5 Adoption Lags

  • According to the Ramp AI Index, Anthropic was used by 43.5% of U.S. businesses (up 1.1 percentage points month-over-month), continuing to pull away from OpenAI, which stood at 39.7% (+0.23pp), with xAI at 4.0% (+0.94pp). [2]
  • Despite Anthropic's overall enterprise API spending rising sharply, Fable 5 accounted for only approximately 6% of Anthropic's enterprise API token purchases and 11.4% of spend in its first month — indicating businesses are largely sticking with older Claude models rather than upgrading to Fable 5. [3]
  • The Ramp data carries a caveat noted in the sources: the sample is drawn from Ramp's own enterprise customer base, which skews toward tech companies, meaning the figures may not be representative of the broader market. [3]
  • Anthropic reduced Claude Code's system prompt by 80% because the Fable 5 models "want a smaller system prompt," reflecting a technical change in how the newest model generation is optimized for agentic coding use cases. [1]
AnthropicOpenAIxAIRampFable 5ClaudeClaude CodeRamp AI Indexenterprise AI adoptionAPI spendingmarket shareagentic codingmodel upgrade ratesAI index benchmarkingsystem prompt optimizationbusiness AI usage trends
Sources
  1. [1]the-decoder.com — Anthropic says it cut 80 percent of Claude Code's system prompt because Fable 5 models "want a smaller system prompt"★news
  2. [2]wallstengine — Ramp’s July AI Index: Anthropic was used by 43.5% of U.S. businesses, up 1.1pp MoM. OpenAI: 39.7%, +0.23pp xAI: 4.0%, +0.94pp Anthropic’s Fatwitter
  3. [3]WWTLitee — Fable 5 上线首月企业采用率才6%，这数据真让人意外！ 根据 Ramp 数据，Fable 5 约占 Anthropic 企业 API Token 采购量的6%，支出占11.4% 但数据样本来自 Ramp 自有企业，偏向科技公司，不能直接推广到整个市场 这个采用率看起来不高，twitter
11.11

AI Drug Discovery Progress Assessed as Clinically Limited Despite Hype

  • A *Nature Reviews Drug Discovery* Perspective finds that although a wide variety of AI methods have been developed, applied, and benchmarked in drug discovery over the past decade, evidence of their clinically relevant impact is "disappointingly limited." The authors identify key reasons including insufficient focus on clinical translation during model development, difficulties applying AI algorithms to conditional life science data, underspecification of computational models for real-world use cases, and a "technology push" dynamic outpacing "science pull." [1][4]
  • According to QZ, AI systems can screen billions of virtual molecular compounds in days — a process that historically required physical screens of chemical libraries and took years — by training on databases of known molecular structures and biological effects to flag promising candidates before any lab work begins. Protein structure prediction is cited as a second significant application area, with the traditional full drug-discovery pipeline (target identification through clinical trials) previously taking a decade or more. [2]
  • BioSpace reports that an emerging group of AI-centric companies is driving a "fail-fast" drug development model, attracting substantial tech investment into biopharma and pressuring legacy drugmakers to adopt advanced machine learning. GATC Health's chief business officer Tyrone Lam argues the mandate for AI must go beyond discovering molecules faster, toward moving risk management from the end of the drug development cycle to the front — including pre-clinical optimization and better clinical trial design. [3]
Nature Reviews Drug DiscoveryGATC HealthTyrone LamBioSpaceQZAI in drug discoveryclinical translation gapvirtual molecular screeningprotein structure predictionfail-fast drug developmentmachine learning in biopharmarisk management in clinical trialstechnology push vs science pullpre-clinical optimizationcomputational model underspecification
Sources
  1. [1]NatRevDrugDisc — Artificial intelligence in drug discovery — what it is, where we stand and the path forward https://t.co/WPql3fM9bi https://t.co/h0O74sMlUD twitter
  2. [2]qz — AI is rewriting the rules of drug discovery: What once took a decade in the lab can now happen in months, thanks to AI systems screening biltwitter
  3. [3]biospace — AI is quickly becoming a central force in drug development, from powering pharma engines to permanently rewiring the capital markets. https:twitter
  4. [4]salilkallianpur — AI in drug discovery has attracted increasing interest. It is now time for a critical review of progress: where did we advance, where are wetwitter
11.12

Novo Nordisk and AWS Partner on AI Drug Discovery Infrastructure

  • Novo Nordisk and AWS announced a strategic partnership in which AWS will become Novo Nordisk's preferred cloud and strategic AI partner, with the collaboration aimed at accelerating drug discovery and expanding AI use across R&D, commercial operations, and IT; financial terms were not disclosed. [2][3]
  • As a concrete output of the deal, the two companies are establishing a co-innovation hub at Novo Nordisk's London facility, where AWS engineers, AI specialists, and scientists will work directly with Novo Nordisk research teams to analyze scientific data and help accelerate drug discovery. [2][3]
  • The partnership builds on an existing relationship: more than 25,000 Novo Nordisk employees already use AWS-powered AI tools for productivity, and the company previously used AWS technology to develop a generative AI chatbot for information retrieval and document drafting. [2][3]
  • Separately, AWS launched a $1 billion Forward Deployed Engineering unit specifically designed to accelerate enterprise AI adoption, signaling a broader infrastructure-scale commitment to deploying AI within large organizations. [1]
Novo NordiskAWSAmazoncloud and AI strategic partnershipdrug discovery accelerationco-innovation hubenterprise AI adoptionR&D and commercial AI integrationforward deployed engineeringgenerative AI for productivitypharmaceutical-tech collaboration
Sources
  1. [1]MarketScale — AWS launches $1 billion Forward Deployed Engineering unit to accelerate enterprise AI adoption★news
  2. [2]BioNJ_Org — Novo Nordisk, AWS partner on AI drug discovery https://t.co/D3O0O3pd3e #DrugDiscovery #AIDrugDiscovery #PharmaInnovation #HealthcareAI #Novo★twitter
  3. [3]BioNJ_Org — Novo Nordisk, AWS partner on AI drug discovery https://t.co/D3O0O3pd3e #DrugDiscovery #AIDrugDiscovery #PharmaInnovation #HealthcareAI #Novotwitter
11.13

Anthropic expands Claude enterprise reach via corps, design, and education

  • Anthropic launched Claude Corps, a $150 million initiative to train and deploy 1,000 young workers into 400 U.S. nonprofits, embedding AI capabilities into the nonprofit sector at scale. [1][3]
  • VentureBeat reported Anthropic shipped a major Claude Design overhaul featuring design system imports, code round-trips, and a fix for its token-burning problem, addressing key developer pain points in the product. [2]
  • Anthropic announced a new feature for Claude Code users, described by VentureBeat as potentially game-changing, though the specific capability was not fully detailed in the available source text. [4]
  • Indagari's analysis of 28 million U.S. consumers found Claude's paid subscriber count and revenue rising steadily since January, indicating growing enterprise and consumer adoption. [5]
  • Anthropic released a free 27-minute workshop on prompt engineering for Claude, taught by the engineers who built it, covering how to write more effective prompts without requiring registration or payment. [6][7]
AnthropicClaude CorpsClaudeClaude CodeVentureBeatIndagariAnthropic initiativesClaude product updatesnonprofit AI deploymentprompt engineering educationClaude Code featuressubscriber growthdeveloper toolsAI workforce training
Sources
  1. [1]Forbes — Anthropic has unveiled Claude Corps, a $150 million initiative to train and deploy 1,000 young workers into 400 U.S. nonprofits. The fellows★news
  2. [2]VentureBeat — Anthropic ships major Claude Design overhaul with design system imports, code round-trips, and a fix for its token-burning problem Anthropi★news
  3. [3]Anthropic — Introducing Claude Corps★news
  4. [4]VentureBeat — Credit: VentureBeat made with OpenAI ChatGPT-Images-2.0 Anthropic announced a potentially game-changing new feature for users of Claude Code★news
  5. [5]Ana Maria Constantin — Indagari's analysis of 28 million US consumers shows Claude's paid subscribers and revenue rising steadily since January, while DataCamp rep★news
  6. [6]heyrohitai — 🚨 Anthropic just showed a 27-minute workshop on how to actually write prompts for Claude. Taught by the people who built it. Free. No registtwitter
  7. [7]PrinceKu945 — Anthropic just released a free workshop on how to prompt Claude better. And it's taught by the people who built it. In 24 minutes, you'll letwitter
11.14

IBM and OpenAI Partner to Deploy GPT-5.6 Across Enterprise Operations

  • IBM announced a strategic partnership with OpenAI to deploy AI at scale across core business operations and complex workflows, embedding OpenAI frontier models — specifically GPT-5.6 and products like Codex and ChatGPT Work — into IBM Consulting Advantage, IBM's AI platform that combines AI agents, industry assets, and cybersecurity capabilities to help organizations securely deploy AI at scale and drive business outcomes. [4][2]
  • The partnership includes joint go-to-market initiatives and industry-specific solutions targeting financial services, government, telecommunications, and retail, as well as enterprise domains such as finance, procurement, customer operations, and HR. [4]
  • IBM will deploy forward-deployed units of highly specialized engineers and consultants trained through the OpenAI Partner Network to work directly with clients, accelerating AI implementation across complex business workflows and highly regulated environments. [4]
  • IBM also joined the OpenAI Daybreak Cyber Partner Program, incorporating cybersecurity resilience and defense capabilities as part of the broader enterprise AI deployment effort. [2]
  • OpenAI launched a limited preview of GPT-5.6 for a small group of trusted partners ahead of a broader public release, with the public rollout delayed following a U.S. government review request. [1][3]
IBMOpenAIIBM Consulting AdvantageGPT-5.6CodexChatGPT WorkOpenAI Partner NetworkOpenAI Daybreak Cyber Partner ProgramIBM-OpenAI strategic partnershipenterprise AI deploymentAI agents for business workflowsfrontier model integrationIBM Consulting Advantagecybersecurity resilienceindustry-specific AI solutionsregulated enterprise environmentsgo-to-market AI initiativesGPT-5.6 limited preview
Sources
  1. [1]Engadget — OpenAI Launches A Limited Preview Of GPT-5.6 For A 'Small Group Of Trusted Partners'★news
  2. [2]IBM Newsroom — ARMONK, N.Y., June 22, 2026— IBM (NYSE:IBM) today announced it has joined the OpenAI Daybreak Cyber Partner Program, bringing advanced front★news
  3. [3]thelec.net — OpenAI Unveils GPT-5.6 Preview, Public Release Delayed Following US Government Review Request★news
  4. [4]stocktradernet — IBM Partners with OpenAI to Accelerate Secure AI Deployment for Enterprises Across Core Operations $IBM https://t.co/zQ3zHyE62jtwitter
11.15

Google Gemini 3.7 Flash Launches as Coding and Agents Workhorse

  • Google introduced Gemini 3.7 Flash on August 13 (local time), just ~3 weeks after the release of Gemini 3.6 Flash, positioning it as the company's "most intelligent workhorse model yet for coding and agents." The model delivers substantial benchmark gains over 3.6 Flash in software engineering, web development workflows, and complex knowledge work. [4][5][6][7]
  • Gemini 3.7 Flash is priced at $0.75 per million input tokens and $3.75 per million output tokens, available at half the normal rate through the end of the year. In competitive comparisons, it is characterized as broadly equivalent in capability to Claude Sonnet 5 and GPT-5.6 Terra despite its lower price point. [4][5][8]
  • Google integrated thinking/reasoning capability — previously available only in a separate Gemini 2.5 model — as a native tool directly inside Flash, accessible through the Gemini API. [2]
  • Anthropic released Claude Sonnet 5, which is cited alongside Gemini 3.7 Flash and GPT-5.6 as a competing mid-tier enterprise model in the same release cycle. [1][4][5]
  • TestingCatalog AI News reported that OpenAI may be preparing a GPT-5.6 release, referred to in competitive context alongside the Gemini 3.7 Flash and Claude Sonnet 5 launches. [3][4][5]
GoogleGemini 3.7 FlashGemini 3.6 FlashAnthropicClaude Sonnet 5OpenAIGPT-5.6 TerraGemini APITestingCatalog AI NewsAI model release cyclecoding and agents optimizationreasoning capability integrationmid-tier enterprise modelscompetitive pricingsoftware engineering benchmarksthinking/reasoning toolsAI model comparison
Sources
  1. [1]Anthropic — Introducing Claude Sonnet 5★news
  2. [2]Ana Maria Constantin — The capability that previously required a separate Gemini 2.5 model is now a native tool inside Flash, available through the Gemini API and★news
  3. [3]TestingCatalog AI News — OpenAI might be preparing GPT-5.6 for next week's release★news
  4. [4]madonomori — Google、「Gemini 3.7 Flash」を発表 ～わずか3週間での更新、性能アップで年末まで半額／「Claude Sonnet 5」「GPT-5.6 Terra」相当なの… https://t.co/3XGQf00Vg3 https://t.co/KbWkT4IxQetwitter
  5. [5]madonomori — Google、「Gemini 3.7 Flash」を発表 ～わずか3週間での更新、性能アップで年末まで半額／「Claude Sonnet 5」「GPT-5.6 Terra」相当なの… https://t.co/3XGQf00Vg3 https://t.co/KbWkT4IxQetwitter
  6. [6]gadgetlistings — Google Introduces Gemini 3.7 Flash • Faster, smarter model focused on coding, agents & complex tasks • Major gains in software engineering, twitter
  7. [7]procastx — Google just introduced Gemini 3.7 Flash — and it’s not just about being faster. The new model brings major gains in coding, web development,twitter
  8. [8]taziku_co — GoogleがGemini 3.7 Flashを公開。 3.6 Flashから、コーディング、ツール利用、複数ステップの実行能力を改善。さらに年末までは入力$0.75 / 出力$3.75（100万token） AIエージェントの本流は「十分に賢いモデルを大量に動かす」設計となるかtwitter
11.16

OpenAI Foundation launches $100M AI program for nonprofits and civil society

  • The OpenAI Foundation is launching "AI for Civil Society and Philanthropy," a program pledging $100 million to deploy AI tools for nonprofits, community organizations, and other civil society and philanthropic partners working on urgent real-world challenges. The Foundation describes itself as "the world's largest nonprofit" in positioning this initiative. [1][2][3][4][5][6]
OpenAI FoundationAI for Civil Society and PhilanthropyAI for nonprofitscivil society AI deploymentphilanthropic AI initiativenonprofit technology fundingOpenAI Foundation grant programreal-world AI applicationscommunity organization support
Sources
  1. [1]FoundationOAI — The OpenAI Foundation is launching AI for Civil Society and Philanthropy to help nonprofits, community organizations, and other trusted part★twitter
  2. [2]MTSlive — SITUATION DETECTED: The OpenAI Foundation is launching AI for Civil Society and Philanthropy, a new program to put AI to work on urgent realtwitter
  3. [3]FoundationOAI — The OpenAI Foundation is launching AI for Civil Society and Philanthropy to help nonprofits, community organizations, and other trusted parttwitter
  4. [4]FoundationOAI — The OpenAI Foundation is launching AI for Civil Society and Philanthropy to help nonprofits, community organizations, and other trusted parttwitter
  5. [5]FoundationOAI — The OpenAI Foundation is launching AI for Civil Society and Philanthropy to help nonprofits, community organizations, and other trusted parttwitter
  6. [6]FoundationOAI — The OpenAI Foundation is launching AI for Civil Society and Philanthropy to help nonprofits, community organizations, and other trusted parttwitter
11.17

Matic Robots launches voice and gesture control update for home robot

  • Matic Robots has released "Cues," a voice and gesture control update for its home-cleaning robot, allowing users to direct the robot using natural speech paired with pointing gestures; the system is powered by 5 RGB-IR cameras and on-device NVIDIA processing, with all visual data processed locally without sending footage to the cloud. [1][2]
  • The Cues update is being highlighted in robotics and AI circles as a milestone in on-device visual perception for consumer robots, with Matic's co-founder Navneet Dalal — known for publishing the influential 2005 "Histograms of Oriented Gradients" (HOG) paper on human detection — credited as a key figure behind the robot's vision capabilities. [3][4]
  • Matic Robots has raised $115M in total funding in the course of developing what it describes as the world's first intuitive home robot, with the Cues launch representing a major product milestone after nine years of development. [1]
Matic RobotsCuesNVIDIANavneet DalalHistograms of Oriented Gradientshome-cleaning robotvoice and gesture controlon-device visual perceptionprivacy-first local processingconsumer robotics milestonemultimodal human-robot interactionRGB-IR camera sensing
Sources
  1. [1]maticrobots — Introducing Cues: Voice & Gesture control for Matic We raised $115M and spent 9 years to make the world’s first intuitive home robot Say you★twitter
  2. [2]testingcatalog — Matic has released an update for its home-cleaning robot, allowing it to be controlled by voice and gestures! Users can use voice paired wittwitter
  3. [3]alex_prompter — The guy who taught computers to see humans just built a robot for your house. In 2005, Navneet Dalal published a paper called Histograms of twitter
  4. [4]andrew_edgeai — Shoutout to @maticrobots for the masterclass on UX and building a robot people want to use! They built an experience around the most naturaltwitter
11.18

Apple trains China-specific LLM in partnership with Alibaba

  • Apple has reportedly trained a proprietary large language model specifically for the Chinese market, distinct from its global Apple Intelligence stack, in a departure from its usual strategy of relying on external AI partners in each region. [1][2][3]
  • Alibaba served as a key development partner on the model, with Apple leveraging Chinese cloud and AI infrastructure to navigate Chinese AI regulations and compete locally. [2][3][4]
AppleAlibabaApple IntelligenceApple LLM for Chinaproprietary language modelChinese AI marketAlibaba partnershipChinese AI regulationsregional AI strategycloud infrastructurelocalization
Sources
  1. [1]japantimes — Apple has trained a large language model specifically for the China market, sources say, a departure from the iPhone maker's strategy of rel★twitter
  2. [2]tldrnewsletter — Apple has trained a large language model of its own for the China market with Alibaba's support, Reuters reports, a shift from its reliance twitter
  3. [3]MarioNawfal — 🇺🇸🇨🇳 Apple has trained its own AI model specifically for China, with help from Chinese tech giant Alibaba. It's a remarkable workaround to Ctwitter
  4. [4]MTSlive — SITUATION DETECTED: Apple has quietly trained a custom, proprietary large language model specifically tailored for China. The model was devetwitter
11.19

Oracle and Google expand Gemini AI integration across enterprise applications

  • Oracle and Google Cloud expanded their partnership to bring Google's Gemini models — specifically Gemini 3.1 Flash-Lite and 3.5 Flash — into Oracle AI Agent Studio for Fusion Applications, a platform for building, connecting, and running AI automation and agentic applications using reusable agents. Oracle also plans to use Gemini models for embedded AI use cases in Oracle Fusion Applications and Oracle NetSuite, targeting workflow automation, faster decisions, and mission-critical execution. [4][5][6][7][8]
  • Google Cloud's Gemini Enterprise Agent (formerly known as Vertex AI Agent Builder) is Google's end-to-end enterprise platform for building, grounding, and scaling multimodal AI agents; it natively uses Gemini models including Gemini 1.5 Pro and 1.5 Flash with context windows exceeding 2 million tokens, and integrates with over 200 open and third-party models via Model Garden. The platform supports use cases including customer service automation, internal knowledge assistants, regulatory compliance auditing, and multimodal media analysis, with data processed within the platform never used to train base foundation models. [3]
  • GM announced it is bringing Google Gemini AI to Cadillac and Corvette vehicles, extending enterprise-grade AI into the automotive product lineup. [1]
  • Oracle outlined significant risks in its heavy bet on AI infrastructure and services, according to The Register's reporting on the company's risk disclosures. [2]
OracleGoogle CloudGeminiOracle AI Agent StudioOracle Fusion ApplicationsOracle NetSuiteGemini Enterprise AgentVertex AI Agent BuilderModel GardenGMCadillacCorvetteOracle-Google Cloud partnershipGemini models enterprise integrationAI agent platformsagentic workflow automationautomotive AI deploymententerprise foundation model integrationmultimodal AI agentsAI infrastructure risk disclosure
Sources
  1. [1]GM Authority — GM Announces Google Gemini AI For Cadillac And Corvette★news
  2. [2]The Register — Oracle outlines all the ways it could lose the farm it bet on AI★news
  3. [3]AilabMobi — Gemini Enterprise Agent https://t.co/5akSxqUXTNtwitter
  4. [4]tarunb_us — Oracle is adding #Google Gemini models to #Oracle AI #Agent Studio for #Fusion Applications to give customers and partners more choice,expantwitter
  5. [5]steveking4 — Oracle is adding Google Gemini models to Oracle AI Agent Studio for Fusion Applications to give customers and partners more choice, expandedtwitter
  6. [6]rahulkurkute10 — Oracle is adding Google Gemini models to Oracle AI Agent Studio for Fusion Applications to give customers and partners more choice, expandedtwitter
  7. [7]KhanafyMahmoud — Oracle is adding Google Gemini models to Oracle AI Agent Studio for Fusion Applications to give customers and partners more choice, expandedtwitter
  8. [8]gclaessen — Oracle is adding Google Gemini models to Oracle AI Agent Studio for Fusion Applications to give customers and partners more choice, expandedtwitter
11.20

ChatGPT web traffic share drops 23 points as rivals gain ground

  • According to Similarweb's generative AI website traffic share data, ChatGPT's share fell from 77.5% to 54.8% over 12 months, a decline of roughly 23 percentage points, while Gemini grew from 9.4% to a larger share and DeepSeek, Grok, Copilot, and Claude also held portions of the remaining traffic. [1][2][3][4][5]
  • Perplexity AI's share of AI referral traffic slipped from 1.9% to 1.0% over the same 12-month period tracked by Similarweb. [1][2][3][4][5]
SimilarwebChatGPTGeminiDeepSeekGrokCopilotClaudePerplexity AIgenerative AI website trafficChatGPT market share declineAI chatbot competitionAI referral trafficsearch and AI traffic trends
Sources
  1. [1]Similarweb — Gen AI website traffic share update - July 2026 🗓️ 12 months ago: ChatGPT: 77.5% Gemini: 9.4% DeepSeek: 4.2% Grok: 3.1% Copilot: 2.0% Claudetwitter
  2. [2]Similarweb — Gen AI website traffic share update - July 2026 🗓️ 12 months ago: ChatGPT: 77.5% Gemini: 9.4% DeepSeek: 4.2% Grok: 3.1% Copilot: 2.0% Claudetwitter
  3. [3]Similarweb — Gen AI website traffic share update - July 2026 🗓️ 12 months ago: ChatGPT: 77.5% Gemini: 9.4% DeepSeek: 4.2% Grok: 3.1% Copilot: 2.0% Claudetwitter
  4. [4]Similarweb — Gen AI website traffic share update - July 2026 🗓️ 12 months ago: ChatGPT: 77.5% Gemini: 9.4% DeepSeek: 4.2% Grok: 3.1% Copilot: 2.0% Claudetwitter
  5. [5]Similarweb — Gen AI website traffic share update - July 2026 🗓️ 12 months ago: ChatGPT: 77.5% Gemini: 9.4% DeepSeek: 4.2% Grok: 3.1% Copilot: 2.0% Claudetwitter
11.21

OpenAI CRO Denise Dresser departs amid wave of senior executive exits

  • Denise Dresser, who joined OpenAI as Chief Revenue Officer in December after serving as CEO of Slack, announced she is leaving the company in the "coming weeks" to "pursue other opportunities," according to a note she posted to LinkedIn. She is being replaced by Dali Rajic, formerly President and COO of cybersecurity company Wiz, who OpenAI says will "build the revenue operating system needed to scale for this next" phase. [3][4]
  • Dresser's departure is the second senior exit within days: Brad Lightcap, a top OpenAI executive and friend of CEO Sam Altman who had been in a "special projects" role after transitioning from COO, announced he was leaving to "start something new" after eight years at the company. Former AGI chief Fidji Simo and former CMO Kate Rouch also recently stepped down. [3][4]
  • The New York Post notes that OpenAI has been scrambling to catch up to rival Anthropic in enterprise sales — a segment that sent Anthropic revenue surging — and that Dresser's exit creates a particular void because she had been brought on partly to absorb some of Lightcap's duties and had brought enterprise experience to the company. [4]
  • Noam Shazeer, Google's Vice President of Engineering and co-lead of its Gemini AI models, announced he was leaving Google, with TechCrunch reporting OpenAI is bringing on big names including Shazeer ahead of its anticipated public debut. [1][2]
OpenAIAnthropicGoogleWizSlackDenise DresserDali RajicBrad LightcapFidji SimoSam AltmanNoam ShazeerGeminiOpenAI executive departuresenterprise sales competitionC-suite leadership turnoverAnthropic revenue growthOpenAI IPO preparationsenior talent acquisitionchief revenue officer successionAI company organizational changes
Sources
  1. [1]TechCrunch — OpenAI is bringing on some big names to the team in the lead-up to its public debut: Google DeepMind AI legend Noam Shazeer and former Trump★news
  2. [2]CNBC — Google's vice president of engineering and a co-lead of its Gemini AI models Noam Shazeer announced Wednesday that he was leaving the compan★news
  3. [3]verge — Chief revenue officer Denise Dresser announced she is leaving OpenAI. Two days ago, Brad Lightcap also said he’d be leaving. https://t.co/WUtwitter
  4. [4]nypost — OpenAI revenue chief Denise Dresser departs - latest high-level exit ahead of blockbuster IPO https://t.co/thM9wpvFTr https://t.co/1uZLff5ettwitter
11.22

SK Telecom Plans Massive AI Data Center Expansion; KOSPI Surges on Chip Rally

  • SK Telecom has announced plans to build out 15 GW of AI data center capacity across South Korea by 2035, representing a total investment of approximately $653 billion. [1][2][3]
  • South Korea's KOSPI index surged approximately 22% over 10 days, entering a technical bull market driven by an AI-fueled semiconductor rally, with the index reaching approximately 6,977.94. This followed a bear market the prior month, reversing a historic 23% rout. [4][6][7]
  • Semiconductor names led the KOSPI rally, with Samsung Electronics among the AI memory leaders that investors piled into, and South Korea's semiconductor exports reaching a historic $21.3 billion record. [5][7]
SK TelecomKOSPISamsung ElectronicsSouth KoreaAI data center investmentsemiconductor rallySouth Korea tech stocksAI memory chipssemiconductor exportsinfrastructure buildoutbull market recovery
Sources
  1. [1]W.Media — SK Telecom plans 15 GW AI data center expansion across South Korea★news
  2. [2]The Korea Economic Daily Global Edition — SK Telecom plans $653 billion AI data center buildout in Korea★news
  3. [3]The Korea Herald — SK Telecom targets 15 GW AI data center capacity by 2035★news
  4. [4]EconomicTimes — The chip rally is charging back with a vengeance | South Korea's Kospi index is on the verge of a technical bull market after rising 22% frotwitter
  5. [5]Bittopia_ — Good Evening Everyone..! 🌏 South Korea’s Kospi jumped about 3.7% as semiconductor names led a fresh AI-driven rally. The move is another remtwitter
  6. [6]MSBIntel — South Korean stocks surge 22% in 10 days, entering technical bull market as AI chip rally reverses July's historic 23% rout. https://t.co/Gktwitter
  7. [7]YUBIT_Exchange — Korea’s chip rally just hit Day 5. 🔥 | Aug 14 $KOSPI surged 2.42% to 6,977.94 as investors piled into AI memory leaders. 📈 Samsung Electronitwitter
11.23

xAI Releases Grok 4.6 and Grok Imagine Video 1.5 for Enterprise Use

  • xAI released Grok 4.6 on August 12, 2026, positioning it as a major upgrade over Grok 4.5, with enhanced capabilities in long-running multi-step AI agents, coding, complex knowledge work, and visual content creation; the model is priced at $2/million input tokens and $6/million output tokens — the same price as its predecessor. [8][9][10]
  • According to xAI's benchmarks, Grok 4.6 matches GPT-5.6 Sol on the AA Intelligence benchmark score of 61, and the model has been integrated into Cursor AI as part of a strategy to compete with Anthropic in the coding space. [6][7][3]
  • xAI shipped the Grok Build Plugin Marketplace at launch with integrations including MongoDB, Vercel, Sentry, Chrome DevTools, Cloudflare, and Superpowers plugins, expanding Grok's enterprise developer ecosystem. [2]
  • xAI released Grok Imagine Video 1.5, a faster AI video generation model featuring improved audio and physics simulation, with the model rendering video in approximately 25 seconds. [1][5][4]
xAIGrok 4.6GPT-5.6 SolCursor AIAnthropicMongoDBVercelSentryCloudflareGrok Imagine Video 1.5AA IntelligenceGrok Build Plugin MarketplaceAI model releaseagentic AI capabilitiescoding assistant competitionAI video generationdeveloper plugin ecosystementerprise AI toolsbenchmark comparisonmulti-step AI agents
Sources
  1. [1]xAI — Grok Imagine Video 1.5★news
  2. [2]MarkTechPost — xAI Ships Grok Build Plugin Marketplace With MongoDB, Vercel, Sentry, Chrome DevTools, Cloudflare, and Superpowers Plugins at Launch xAI Sh★news
  3. [3]Brian Wang — Home»Artificial intelligence»XAI and Cursor Strategy to Catch Up and Beat Anthropic in Coding Cursor is training its own Opus-class model fr★news
  4. [4]openPR.com — xAI Renders Grok Video in 25 Seconds Yet Pays Creators Nothing as Ruvi (RUVI) Pays Its AI Trainers★news
  5. [5]India Today — Elon Musk's xAI launches faster Grok Imagine Video 1.5 with improved audio and physics★news
  6. [6]kimmonismus — Grok 4.6 released, absolutely insane crazy jump! xAI was cooking! According to xAI’s benchmarks, Grok 4.6 matches GPT-5.6 Sol on the AA Inte★twitter
  7. [7]kimmonismus — Grok 4.6 has become a VERY strong competitor. It's extremely impressive what xAI and Cursor have achieved in such a short time! However, the★twitter
  8. [8]itnavi2022 — 🎉xAIは2026年8月12日、最新AIモデル「Grok 4.6」を発表。Grok 4.5を基盤に、特に長時間・多段階で動くAIエージェント、コーディング、複雑な知識作業、アプリや視覚コンテンツ制作を強化した。調査から実装、改善まで一連の作業を継続し、自らテスト・検証する能力も向twitter
  9. [9]yuru2_Kurage — /／ 8/14 ハナキンキラキラ✨️💰️金曜日 くらげニュース #199 ＼ おはくら〜🤟✨ 【Grok 4.6登場！コスパ最強クラス】 xAIがGrok 4.6をリリース！ 長時間エージェントやコーディング、知識作業に強くなり、GPT-5.6 Solと同じベンチマーク61を達成twitter
  10. [10]yuru2_Kurage — /／ 8/14 ハナキンキラキラ✨️💰️金曜日 くらげニュース #199 ＼ おはくら〜🤟✨ 【Grok 4.6登場！コスパ最強クラス】 xAIがGrok 4.6をリリース！ 長時間エージェントやコーディング、知識作業に強くなり、GPT-5.6 Solと同じベンチマーク61を達成twitter
11.24

Google DeepMind SL2T Sign Language AI Launches on Pixel 11

  • Google DeepMind announced "SL2T" (Sign Language to Text), a multilingual sign language translation AI trained on over 50 sign languages and more than 100,000 hours of data, to be integrated into Gboard and Live Transcribe on the Pixel 11, initially supporting American Sign Language (ASL) to English text conversion. [1][2][3]
  • SL2T uses on-device MediaPipe Holistic to detect body landmark coordinates — covering hands, arms, torso, head, and face — and transmits only those geometric coordinates to the translation model rather than raw camera video, which is immediately discarded as a privacy measure. [3]
  • The model translates directly from body landmark sequences to text without using "gloss" intermediate representations, a method that previously required costly expert annotation and limited available training data; Google DeepMind reports that joint training across multiple sign languages, dialects, and proficiency levels outperformed single-language models in internal experiments. [3]
  • On the FLEURS-ASL sd-test benchmark, SL2T achieved a zero-shot score of 70 BLEURT, which Google DeepMind states substantially exceeds previously reported scores for sign language translation. [3]
  • Intended use cases for SL2T on Pixel 11 include web search, composing messages and documents, querying Gemini, and facilitating face-to-face conversations via Live Transcribe, positioning sign language as a direct input interface equivalent to voice input. [3]
  • Google DeepMind stated plans to expand SL2T support to additional devices and sign languages beyond the initial ASL-to-English deployment. [3]
Google DeepMindSL2TGboardLive TranscribePixel 11MediaPipe HolisticFLEURS-ASLGeminiAmerican Sign Languagesign language translation AImultilingual sign language recognitionon-device privacy-preserving inferencebody landmark detectiongloss-free translationzero-shot benchmark performanceaccessibility technologyreal-time transcriptioncross-lingual transfer learning
Sources
  1. [1]masahirochaen — 【速報】Google DeepMindが手話をそのまま文字にするAI「SL2T」を公開。Pixel 11から搭載。 ・ASL→英語をGboardに手話で直接入力 ・Live Transcribeで手話のまま会話 ・50以上の手話10万時間で学習 手話の翻訳は大義もあり画期的な技術twitter
  2. [2]heiga_zen — 【手話AIをユーザの手元へ：Pixel 11のGboardとLive Transcribeに搭載された "SL2T"】 "Made by Google" にて、Google DeepMindとAndroidチームによる手話翻訳モデル「SL2T」の製品展開が発表されました。ユーザーtwitter
  3. [3]innovaTopia_JP — Google DeepMind「SL2T」｜手話AIがPixel 11の入力機能へ、ASLを英語テキストに変換 https://t.co/Qz4GCeQHFc キーボードや声を前提に作られてきた道具に、もう一つの自然な伝え方が加わろうとしている。機械が人に合わせる方へ動いている。twitter
11.25

Google DeepMind leadership overhaul as Hassabis exits CEO role

  • Google DeepMind cofounder Demis Hassabis is stepping down as CEO and moving to a chairman role, with the transition announced jointly by Google CEO Sundar Pichai and Hassabis to employees at the end of the day in London; simultaneously, longtime chief scientist Jeff Dean and several other prominent researchers announced they were leaving the company. [4][5]
  • According to Fortune's reporting based on accounts from three current and one former DeepMind employee, internal tensions were already simmering before the announcement, citing delayed models — specifically the Gemini 3.5 Pro model, which had missed three release deadlines — high-profile talent exits, and a public dispute with staff over controversial defense contracts. [4][5]
  • DeepMind staff in the London office were described as "very mixed" on the leadership reshuffle, with some questioning whether Hassabis's chairman role would ultimately result in him being phased out entirely, and others wondering if it signaled a power shift from the London lab to Google's Mountain View headquarters. [4][5]
  • Nobel Prize-winning AlphaFold researcher John Jumper, who shared the chemistry Nobel with Demis Hassabis, announced he is leaving Google DeepMind for Anthropic after nearly nine years at the lab. [1][2][3]
Google DeepMindDemis HassabisSundar PichaiJeff DeanGemini 3.5 ProJohn JumperAnthropicAlphaFoldGoogleMountain ViewGoogle DeepMind leadership transitionCEO resignationAI talent exodusinternal organizational tensionsdefense contracts controversymodel release delaysLondon vs Mountain View power shiftNobel Prize researcher departure
Sources
  1. [1]Business Insider — AlphaFold pioneer who won a Nobel Prize alongside Demis Hassabis leaves Google DeepMind for Anthropic★news
  2. [2]Anthony Ha — John Jumper, who shared a recent Nobel Prize in chemistry, announced Friday that he’s making the leap to Anthropic after “nearly 9 years” at★news
  3. [3]CNBC — John Jumper to leave Google DeepMind for Anthropic★news
  4. [4]FortuneMagazine — Google DeepMind cofounder Demis Hassabis is stepping down as CEO and moving to a chairman role. Fortune’s @beafreyanolan explains how stalletwitter
  5. [5]FortuneMagazine — Google DeepMind cofounder Demis Hassabis is stepping down as CEO and moving to a chairman role. Fortune’s @beafreyanolan explains how stalletwitter
11.26

Vals AI raises $40M Series A to build real-world AI evaluation infrastructure

  • Vals AI announced a $40M Series A at a $400M valuation, led by Andreessen Horowitz (a16z), with participation from existing investors 8VC and Pear VC. [1][2]
  • The core investment thesis, as articulated by a16z investor Jennifer Li, is that frontier models increasingly ace standard academic benchmarks — which have become saturated, leak into training corpora, or are explicitly optimized against — yet still fail at the complex, multi-step tasks that matter in real-world enterprise deployments such as analyzing financial documents, resolving legal cases, writing software, and navigating enterprise workflows. [3]
  • Vals AI's approach is to work with domain experts across fields to turn real workflows into rigorous benchmarks, then build automated grading systems that evaluate final work product — testing models on tasks people actually need done rather than contrived academic exams. [3]
  • The company is specifically positioned to address rising stakes as AI models shift from question-answering to autonomous agentic work operating over hours and days, where a poor model choice can cost thousands of dollars in tokens plus time and customer satisfaction. [3]
Vals AIAndreessen Horowitz8VCPear VCJennifer LiAI benchmarkingenterprise AI evaluationagentic AI systemsbenchmark saturationreal-world workflow assessmentfrontier model limitationsautomated gradingdomain-specific benchmarksSeries A fundingmodel selection
Sources
  1. [1]ValsAI — Today we're announcing our $40M Series A at a $400M valuation, led by @a16z , with participation from existing investors @8vc, @pearvc, and ★twitter
  2. [2]a16z — We're thrilled to invest in Vals. A frontier model can look brilliant on a leaderboard and still struggle with the messy work that actually ★twitter
  3. [3]JenniferHli — 1/ We’re thrilled to announce our investment in @ValsAI. AI has gotten very good at taking tests and winning trophies. However, it's still htwitter
11.27

Elon Musk predicts AI will represent 99% of SpaceX's value

  • Elon Musk publicly stated that within four to five years, AI will constitute 99% of SpaceX's value, adding that "the value of SpaceX will be some astronomical number." [1][2][3][4][6]
  • Musk also discussed a "Star Mind" factory, describing it as located next to SpaceX and central to the company's AI direction, though specific operational or product details were not elaborated upon in the sources. [5]
Elon MuskSpaceXStar MindAI valuationSpaceX AI strategyfuture of AI in aerospaceElon Musk predictionsAI-driven company valueStar Mind factory
Sources
  1. [1]cb_doge — "Long term, probably in four or five years, AI will be 99% of the value of SpaceX. I'd say five years for sure, AI will be 99% the value of ★twitter
  2. [2]niccruzpatane — Elon Musk says in 4-5 years, AI will be 99% of the value of @SpaceX. "And the value of SpaceX will be some astronomical number." Hold onto y★twitter
  3. [3]cb_doge — "Long term, probably in four or five years, AI will be 99% of the value of SpaceX. I'd say five years for sure, AI will be 99% the value of ★twitter
  4. [4]niccruzpatane — Elon Musk says in 4-5 years, AI will be 99% of the value of @SpaceX. "And the value of SpaceX will be some astronomical number." Hold onto ytwitter
  5. [5]teslaownersSV — Elon Musk on the future of Star Mind. Musk says the Star Mind factory will be located right next to SpaceX. But his bigger prediction is abotwitter
  6. [6]CosmosEuropa — 🚨 Elon Musk says AI could become 99% of SpaceX's value within 5 years. His prediction: “Probably in four or five years, AI will be 99% of thtwitter
11.28

OpenAI Runs Developer Credit Giveaway and Codex Quota Reset Test

  • OpenAI is offering open-source developers up to $1,200 in ChatGPT Pro credits to use its Codex coding tool; eligibility requires maintaining at least one public GitHub repository and completing an application form. [1][4]
  • OpenAI is internally testing a paid quota-reset feature for Codex, in which Plus users who have exhausted their monthly usage limit see a "pay $8 to reset immediately" button, allowing them to restore their quota without upgrading to a higher tier or waiting for the next billing cycle. The reset pricing and coverage for Pro-tier users have not yet been disclosed. [2][3]
  • Some sources note that reporting has conflated two separate OpenAI products, with the $8 reset applying specifically to Codex usage quotas rather than general ChatGPT limits, according to clarification found in OpenAI's official documentation. [3]
OpenAIChatGPTCodexGitHubopen-source developer creditsCodex coding toolusage quota resetpaid quota restorationChatGPT Plus subscriptionOpenAI pricing modelcoding assistant access
Sources
  1. [1]sabir_huss50540 — OpenAI is giving open-source developers up to $1,200 in ChatGPT Pro for free. Barely anyone is talking about it. If you maintain an open-soutwitter
  2. [2]GeekCatX — OpenAI 开始内测 Codex「付费重置」能力。 一位 Reddit 用户在 $20/月的 Plus 额度耗尽后，看到了「支付 $8，立即重置」的按钮。 不用升级，也不用等下周。 再付一次钱，原地满血。 Pro 套餐的重置价格和覆盖范围目前还不清楚。 假设 Codex 每周有twitter
  3. [3]Chong19171314 — 等等，8美元就能让 ChatGPT 一周额度“原地复活”？ 如果真是这样，Plus 用户还升什么 Pro。 但我翻完 OpenAI 官方说明，发现这条消息把 ChatGPT 和 Codex 混在了一起。 OpenAI 现在开放的是：部分 Plus、Pro 用户在 Codex 套餐twitter
  4. [4]sabir_huss50540 — OpenAI is giving open-source developers up to $1,200 in ChatGPT Pro for free. Barely anyone is talking about it. If you maintain an open-soutwitter
11.29

AI coding startup Cursor plans first India office amid SpaceX acquisition

  • Cursor, an AI coding startup, is planning to open its first physical office in India by the end of the year, driven by rapid user growth in the region. The expansion coincides with a reported pending acquisition of Cursor by SpaceX. [2][3][4]
  • Brian Wang reports that xAI and Cursor are pursuing a strategy to compete with and surpass Anthropic in coding, with Cursor training its own Opus-class model as part of this effort. [1]
CursorSpaceXxAIAnthropicBrian WangIndiaOpusAI coding toolsstartup expansionIndia office openingacquisition dealmodel trainingcoding competitionfrontier model development
Sources
  1. [1]Brian Wang — Home»Artificial intelligence»XAI and Cursor Strategy to Catch Up and Beat Anthropic in Coding Cursor is training its own Opus-class model fr★news
  2. [2]IndianTechGuide — 🚨 Elon Musk-owned Cursor plans to set up its first office in India by the end of this year. https://t.co/sCOVQL2g7gtwitter
  3. [3]chandrarsrikant — 🚨Cursor to open first India office by year-end amid rapid user growth Artificial intelligence (AI) coding startup Cursor, which is set to betwitter
  4. [4]YourStoryCo — AI coding startup @cursor_ai, set to be acquired by @elonmusk's @SpaceX — plans to open its first physical office in India before the end oftwitter
11.30

Oracle promotes APEX AI Application Generator for enterprise app development

  • Oracle is marketing its APEX AI Application Generator as a solution for building enterprise-class applications, emphasizing that it brings the speed of AI to application development while incorporating governance, security, and control features intended to meet enterprise requirements. [2][3][4][5][6]
  • Oracle developer Leon van Zyl is featured in a demonstration showing how to build an agentic application from start to finish using the platform, which Oracle describes as combining AI coding agents, skills, and CLIs. [7]
  • The Register reports that Oracle has outlined multiple risk factors related to its bet on AI, flagging ways the company could face significant losses if its AI strategy does not succeed. [1]
OracleAPEX AI Application GeneratorLeon van ZylThe Registerenterprise application developmentAI coding agentsagentic applicationsOracle APEXAI governance and securityAI strategy risk factorslow-code AI platform
Sources
  1. [1]The Register — Oracle outlines all the ways it could lose the farm it bet on AI★news
  2. [2]OracleAPEX — AI can build applications. Oracle APEX AI Application Generator helps build enterprise applications—with governance, security, and control b★twitter
  3. [3]OracleDatabase — AI can generate code. But can it generate enterprise-class applications? Oracle APEX AI Application Generator brings the speed of AI to a gotwitter
  4. [4]OracleExadata — AI can build applications. Oracle APEX AI Application Generator helps build enterprise applications—with governance, security, and control btwitter
  5. [5]OracleDatabase — AI can build applications. Oracle APEX AI Application Generator helps build enterprise applications—with governance, security, and control btwitter
  6. [6]OracleExadata — AI can generate code. But can it generate enterprise-class applications? Oracle APEX AI Application Generator brings the speed of AI to a gotwitter
  7. [7]DermotEustace — Watch Leon van Zyl build an agentic application from start to finish and learn how Oracle is bringing AI coding agents, skills, CLIs, and fatwitter
11.31

AI Tools Used to Design Personalized mRNA Cancer Vaccine for Dog

  • A dog named Rosie was given months to live after chemotherapy and immunotherapy failed; her owner used ChatGPT and computational genomics to design a personalized mRNA cancer vaccine, which reportedly worked, and the effort has since become a Y Combinator startup. [1][3]
  • According to a tweet referencing the AlphaFold team's dissolution, Google disbanded its dedicated AlphaFold research team despite the technology's role in landmark scientific achievements, including its reported use in designing the mRNA vaccine for Rosie. [2][3]
ChatGPTY CombinatorAlphaFoldGooglepersonalized mRNA cancer vaccinecomputational genomicsveterinary oncologyAI-driven drug designAlphaFold team dissolutionAI startup formationcancer immunotherapy failure
Sources
  1. [1]paul_conyngham — My dog Rosie was given months to live. Chemo and immunotherapy failed I used chatgpt + computational genomics to design a personalised mRNA ★twitter
  2. [2]acidrender — Google just killed its dedicated AlphaFold research team. Remember the guy who designed an mRNA vaccine to cure his dog's tumor? Why shut dotwitter
  3. [3]AI_Tower — A man built an mRNA cancer vaccine for his dog using ChatGPT and AlphaFold. It worked. Now it's a Y Combinator startup. → Rosie's mast cell twitter
11.32

IBM and Together AI Sign $240M Deal for NVIDIA-Powered AI Inference Cluster

  • IBM and Together AI signed a $240M multi-year agreement to deploy a large-scale open-source AI inference cluster on IBM Cloud, described as the first large-scale NVIDIA-powered open-source inference deployment on that platform. [1][3][5]
  • The infrastructure will be built on NVIDIA HGX B300 systems and NVIDIA Spectrum networking, with Together AI leveraging these components to scale enterprise-grade AI inference on IBM Cloud. [4][6]
  • NVIDIA is a collaborating partner in the arrangement, providing the core AI infrastructure — HGX B300 hardware and Spectrum interconnects — that underpins the IBM and Together AI deployment. [2][6]
IBMTogether AINVIDIAIBM CloudHGX B300Spectrumopen-source AI inferenceenterprise AI infrastructurecloud deployment agreementNVIDIA-powered hardwarelarge-scale inference clustermulti-year partnershipGPU networking
Sources
  1. [1]StockSavvyShay — $IBM signs a $240M multi-year deal with Together AI to deploy a large-scale inference cluster on IBM Cloud using $NVDA B300 systems and Spec★twitter
  2. [2]NVIDIAAIInfra — 🤝 @IBM and @togethercompute are working to scale open-source AI inference on IBM Cloud with NVIDIA AI infrastructure. This collaboration heltwitter
  3. [3]wallstengine — $IBM and Together AI signed a multi-year $240M deal to deploy a large-scale AI inference cluster on IBM Cloud using $NVDA HGX B300 systems atwitter
  4. [4]IBMNews — In collaboration with @togethercompute, IBM is scaling open source AI. Together AI will leverage @NVIDIA HGX B300 systems and NVIDIA Spectrutwitter
  5. [5]blckchaindaily — 🚨 $IBM IBM SIGNS $240M MULTI-YEAR DEAL WITH TOGETHER AI TO DEPLOY FIRST LARGE-SCALE INFERENCE CLUSTER ON IBM CLOUD USING NVIDIA $NVDA HGX B3twitter
  6. [6]togethercompute — Together AI is teaming up with @IBM and @nvidia to bring enterprise-grade AI inference to IBM Cloud. A dedicated NVIDIA B300 cluster. Spectrtwitter
11.33

Sergey Brin pressures Google DeepMind to accelerate Gemini amid delays

  • Reuters reported, citing two people familiar with his remarks, that Google co-founder Sergey Brin has urged key AI staff in recent months to go "all-in" on Gemini, addressing hundreds of employees at an April town hall to push Google DeepMind to move faster — prompted in part by Anthropic's preview of its Claude Mythos model. [1][4]
  • Gemini briefly overtook competitors in November but fell behind again after updates from Anthropic and OpenAI; by August, Google had delayed the new flagship Gemini version by two months after internal testing revealed its performance lagged rivals in areas including coding. [2][3][4]
  • Brin, who holds no executive title at Google, has nonetheless informally influenced AI model training and, according to sources, used his co-founder authority to direct resources toward developing "recursive self-improvement" (RSI) — an autonomously self-improving AI capability. [2][3]
  • On August 5, Alphabet announced a leadership overhaul at Google DeepMind: CEO Demis Hassabis stepped aside to become chair with fewer management responsibilities, replaced by deputy Koray Kavukcuoglu; simultaneously, the two original technical co-leads of Gemini departed to co-found a startup. The restructuring also included transferring some Google DeepMind teams into Google proper. [2][3][4]
  • Google declined to respond to Reuters' questions about Brin's recent involvement and the organizational changes at Google DeepMind. [2][3][4]
GoogleSergey BrinGeminiGoogle DeepMindAnthropicClaude MythosOpenAIDemis HassabisKoray KavukcuogluAlphabetGoogle AI leadership shakeupGemini development delaysrecursive self-improvementSergey Brin informal influenceGoogle DeepMind restructuringAI model competitionAnthropic rivalryautonomous self-improving AI
Sources
  1. [1]Reuters — Exclusive: Google co-founder Sergey Brin in recent months has urged key AI staff to go all in on the company's Gemini model as parent Alphab★twitter
  2. [2]itm_aiplus — Google、「Gemini」開発遅延で焦りか 共同創業者ブリン氏が“全力投球”促す https://t.co/S8Cqi6UUNwtwitter
  3. [3]itm_aiplus — Google、「Gemini」開発遅延で焦りか 共同創業者ブリン氏が“全力投球”促す https://t.co/S8Cqi6UUNwtwitter
  4. [4]adaderana — Google co-founder Brin pressures DeepMind to accelerate Gemini as Alphabet reshuffles AI leadership Read more: https://t.co/S507m2vfza #lka twitter
11.34

Anthropic updates Claude in Chrome with cross-platform session continuity

  • Anthropic updated its Claude in Chrome browser extension so the side panel now runs a full Claude Cowork session, replacing the previous isolated browser-only experience. Conversations, skills, and connectors started in Chrome now carry over seamlessly to desktop, web, and mobile apps, with sessions tied to the user's account rather than to any single device or tab. [1][2][3]
  • The key limitation the update addresses was that closing a Chrome tab previously wiped the Claude session entirely, forcing users to restart from scratch; the new architecture persists sessions across platforms. [3][4][5]
  • The Claude Cowork session running in the side panel enables agentic tasks within the browser — such as navigating analytics dashboards, organizing Google Drive files, reading calendars and email, scraping competitor pricing into formatted PowerPoints, and matching calendar attendees to Salesforce contacts — with the user approving actions rather than performing them manually. [2]
AnthropicClaudeClaude CoworkChromeGoogle DriveSalesforcebrowser extension updatepersistent AI sessionscross-platform continuityagentic browser tasksClaude side panelworkflow automationmulti-device session sync
Sources
  1. [1]claudeai — Your Claude in Chrome sessions now carry over to desktop, web, and mobile. Conversations are saved, and your skills and connectors work in t★twitter
  2. [2]claudeai — The side panel now runs the same Claude Cowork session as the desktop, web, and mobile apps. Sessions live with your account, not on any sin★twitter
  3. [3]NovaPovNP — Anthropic has updated its Claude for Chrome extension so the side panel now runs as a full Claude Cowork session. Conversations and multi-st★twitter
  4. [4]AlessioLidozzi — Anthropic has changed what Claude in Chrome is. The browser side panel now runs the same Claude Cowork session used on desktop, web, and mobtwitter
  5. [5]NovaPovNP — Anthropic has updated its Claude for Chrome extension so the side panel now runs as a full Claude Cowork session. Conversations and multi-sttwitter
11.35

River AI raises $1.1B to build user-owned personal AI

  • River AI, co-founded by former xAI co-founder Igor Babuschkin, raised $1.1 billion in a funding round led by General Catalyst and co-led by AMP Public, with strategic investment from Nvidia and participation from Y Combinator. [1][3][4]
  • River AI's stated mission is to build a "personal AI owned and shaped by each individual," with users owning the hardware it runs on, the data it learns from, and the intelligence itself — positioning the company against centralized AI services. [1][2][4]
River AIIgor BabuschkinxAIGeneral CatalystAMP PublicNvidiaY CombinatorRiver AI funding roundpersonal AI ownershipdecentralized AIuser-owned hardwareAI data ownershipstartup investmentindividual AI autonomy
Sources
  1. [1]river_ai_inc — Today, we're sharing that River AI has raised $1.1 billion, led by @generalcatalyst and @amppublic with strategic investment from @nvidia an★twitter
  2. [2]TheGregYang — "Most importantly, you'll own everything: the hardware it runs on, the data it learns from, and the intelligence itself." 🥰 — quoting @river★twitter
  3. [3]muskonomy — NEWS: Former xAI co-founder Igor Babuschkin raised $1.1 billion for his new AI company, River AI. The round was led by General Catalyst and ★twitter
  4. [4]testingcatalog — River AI raised $1.1 billion 🔥 River AI is aiming to build a "personal AI owned and shaped by each individual." Currently, River AI offers atwitter
11.36

Google Pixel Watch 5 launches with Gemini AI and expanded health features

  • Google announced the Pixel Watch 5, available in 41mm and 45mm sizes, powered by the Qualcomm Snapdragon W5 Gen 2 Accelerated SoC with a Cortex-M55 coprocessor and running Wear OS 7.0. Compared to the prior model, memory was increased by 50%, CPU performance improved by 12%, processing speed by 20%, and storage doubled from 32GB to 64GB. [6]
  • The watch is designed around Gemini Intelligence integration, allowing users to raise their wrist and speak to Gemini to start workouts, set timers, or check reservation information from Gmail. [1][2][3][4][5][6]
  • Health features include a new "Health Guardian" function that provides monthly summaries of "vascular stress," "breathing quality during sleep trends," and "metabolic stress," as well as a Google Health Coach for workout suggestions and a strength training tracker managing sets, rest time, weight, and reps. Sleep stage detection accuracy improved by 15% over the prior model, with a smart alarm that wakes users during light sleep. [6]
  • The display is an Actua 360 AMOLED LTPO panel with 3,000-nit peak brightness, 320ppi resolution, a 1–60Hz refresh rate with always-on support, and a 3D Corning Gorilla Glass 5 cover. Dual-frequency GPS with 3D building models and weather data correction enables route recording at twice the accuracy of the prior model in urban and tree-covered environments. [6]
  • Battery capacity is 332mAh (41mm) and 465mAh (45mm), delivering up to 30 and 40 hours respectively with always-on display enabled, or up to 48 and 72 hours in battery saver mode; rapid charging reaches 50% in approximately 15 minutes for both models. [6]
  • In Japan, the watch launches August 20 with pricing starting at ¥62,800 for the 41mm Bluetooth/Wi-Fi model and reaching ¥84,800 for the 45mm LTE model; online pre-orders opened August 12. [6]
GooglePixel Watch 5Qualcomm Snapdragon W5 Gen 2Cortex-M55Wear OS 7.0GeminiHealth GuardianActua 360Corning Gorilla Glass 5Gmailsmartwatch hardware specsGemini AI integrationhealth monitoring featuresWear OS wearableGPS accuracy improvementAMOLED display technologybattery life and chargingsleep trackingJapan product launchstrength training tracker
Sources
  1. [1]Google — Here’s your first peek at Pixel Watch 5. Designed for Gemini Intelligence and @GoogleHealth, it helps you stay on top of your day with proactwitter
  2. [2]Google — Here’s your first peek at Pixel Watch 5. Designed for Gemini Intelligence and @GoogleHealth, it helps you stay on top of your day with proactwitter
  3. [3]Google — Here’s your first peek at Pixel Watch 5. Designed for Gemini Intelligence and @GoogleHealth, it helps you stay on top of your day with proactwitter
  4. [4]Google — Here’s your first peek at Pixel Watch 5. Designed for Gemini Intelligence and @GoogleHealth, it helps you stay on top of your day with proactwitter
  5. [5]Google — Here’s your first peek at Pixel Watch 5. Designed for Gemini Intelligence and @GoogleHealth, it helps you stay on top of your day with proactwitter
  6. [6]hermita_akiba — Geminiと健康管理機能を強化した「Google Pixel Watch 5」8月20日発売 ➡️Snapdragon W5 Gen 2 AcceleratedとWear OS 7.0を採用 https://t.co/tFLBCHXhdP https://t.co/y8ZQcrtwitter
11.37

Oracle Health launches AI-embedded Patient Portal in U.S.

  • Oracle Health's AI-powered Patient Portal became generally available in the U.S., combining consumer-application ease-of-use with secure integration into Oracle Health's Electronic Health Record (EHR) system, allowing patients to review and share medical records, connect with providers, and manage appointments. [1][2]
  • The embedded AI enables patients to ask questions and receive plain-language explanations of diagnoses, test results, and treatment options, with the portal designed to reduce strain on medical staff handling high call volumes, general inquiries, and intake paperwork. [1][2]
Oracle HealthPatient PortalElectronic Health RecordAI-powered patient portalelectronic health record integrationpatient self-serviceplain-language medical explanationshealthcare AImedical staff workload reductionappointment management
Sources
  1. [1]OracleHealth — Oracle Health Patient Portal is now available in the U.S. Embedded AI helps patients understand their records, manage appointments, and staytwitter
  2. [2]OracleHealth — Oracle Health Patient Portal is now available in the U.S. Embedded AI helps patients understand their records, manage appointments, and staytwitter
11.38

Lovable raises $400M Series C at $13.3B valuation

  • Stockholm-based vibe-coding startup Lovable raised $400 million in a Series C round at a $13.3 billion valuation, led by Menlo Ventures alongside the EQT-managed Scaleup Europe Fund; the round more than doubled the company's valuation in seven months, up from a $6.6 billion valuation when it raised $330 million in December. [1][2][4]
  • Lovable's valuation trajectory has been rapid: it raised a $200 million Series A at a $1.8 billion valuation, then a $330 million round at $6.6 billion, reaching $13.3 billion with this latest raise — approximately a 7x increase in roughly eight months. [7][3]
  • Lovable's ARR has nearly tripled since December, reaching $400 million — up from $300 million a month prior and $200 million at the end of 2025 — with the company disclosing this figure in March. [1][4]
  • Since launching in November 2024, more than 60 million projects have been created on Lovable's platform, and apps built on it now receive 900 million visits. [1][6][4]
  • CEO Anton Osika stated the company will use the funding to evolve Lovable from a software-building tool into a full platform for running a business, with business leaders already using it to "spin up new product lines" and workers using it to "build tools that fit their exact needs." [1][4]
  • Tencent is among the backers in the round, alongside more than a dozen other investors. [5][2]
LovableMenlo VenturesEQTScaleup Europe FundTencentAnton OsikaStockholmvibe-coding startupSeries C funding roundAI software development platformrapid valuation growthno-code app buildingARR growthventure capital investment
Sources
  1. [1]BusinessInsider — Lovable's $400 million raise is the latest sign of investor appetite for vibe coding. https://t.co/7UZ5VXswmAtwitter
  2. [2]IPONewsroom_ — LOVABLE RAISES $400M AT $13.3B VALUATION The Series C was led by Menlo Ventures and the Scaleup Europe Fund, with more than a dozen other intwitter
  3. [3]TradedVC — Lovable, the Swedish vibe-coding startup led by CEO Anton Osika, raised $400 million in Series C funding at a $13.3 billion valuation, nearltwitter
  4. [4]BusinessInsider — Lovable's $400 million raise is the latest sign of investor appetite for vibe coding. https://t.co/7UZ5VXswmAtwitter
  5. [5]Nexorastartups — Swedish vibe-coding startup Lovable raised $400 million at a $13.3 billion valuation (ARR nearing $600 million); Tencent is among the backertwitter
  6. [6]antonosika — Lovable just raised $400M at a $13.3B valuation to build the business that helps build businesses. Apps built on Lovable now get 900M visitstwitter
  7. [7]stretchcloud — Eight months ago Lovable raised a $200M Series A at a $1.8B valuation. Then came a $330M Series B at $6.6B in December 2025. Today: a $400M twitter
11.39

Uber and Pony.ai announce 2,000+ robotaxi deployment across Europe

  • Uber and Chinese autonomous-driving company Pony.ai announced a partnership to deploy more than 2,000 Pony.ai robotaxis across Europe, expanding beyond an existing presence in Zagreb to four additional cities, marking a major international scale-up of Uber's robotaxi operations. [1][5][6]
  • Pony.ai disclosed that its cumulative autonomous-driving distance worldwide has surpassed 100 million miles (the exact figure was truncated in sources), representing a milestone reported alongside the European expansion announcement. [2][3]
  • The global Autonomous Vehicle Market was valued at USD 195.0 billion in 2025 and is projected to reach USD 1,917.1 billion by 2034, growing at a CAGR of 28.9% during 2026–2034, driven by AI and sensor advances; the market is shifting from driver-assistance features toward commercially operated, revenue-generating autonomous fleets, with robotaxi operators moving from pilot programs to metro-wide paid service. [7]
  • China's city-level pilot licensing programs for robotaxi and robotruck operators are active in Beijing, Wuhan, and Shenzhen, and autonomous driving has reportedly launched commercially in Shenzhen, where vehicles are noted to communicate verbally with passengers — a feature distinguishing Chinese deployments from U.S. counterparts. [4][7]
UberPony.aiZagrebBeijingWuhanShenzhenChinarobotaxi deploymentautonomous vehicle market growthUber-Pony.ai partnershipEuropean AV expansionChina autonomous driving pilotscommercial AV fleetsAI and sensor advancesrider-vehicle voice interactionautonomous driving milestones
Sources
  1. [1]Polymarket — JUST IN: Uber & China’s Pony AI plan to deploy more than 2,000 robotaxis across Europe. https://t.co/oTHTlqW8Y2twitter
  2. [2]ChinaScience — Chinese autonomous-driving company Pony. ai said on Monday that its cumulative autonomous-driving distance worldwide has surpassed 100 millitwitter
  3. [3]ChinaScience — Chinese autonomous-driving company Pony. ai said on Monday that its cumulative autonomous-driving distance worldwide has surpassed 100 millitwitter
  4. [4]TheCraneWeekly — 🚕🤖 Autonomous driving has finally launched in Shenzhen! The biggest difference vs. the U.S.? The car actually talks to you. Most China's robtwitter
  5. [5]Vertoveck — Pony AI $PONY and Uber $UBER said Thursday they plan to deploy more than 2,000 Pony AI robotaxis across Europe as the companies expand theirtwitter
  6. [6]asteris_ai — https://t.co/sUL7962kFP and Uber plan more than 2,000 robotaxis across Europe, expanding beyond Zagreb to four more cities. The interesting twitter
  7. [7]KhushiSing80232 — 🚗Autonomous Vehicle Market Advances in AI, sensors, connected vehicle technologies, and autonomous driving systems are accelerating the trantwitter
11.40

Kenya's AI journalism factory floods international editors with pitches

  • A Dispatch investigation traced a surge of AI-generated freelance pitches to Nairobi, finding that a Kenyan platform called Sure Media Agency aggregates freelance opportunities from international news organisations and teaches paid subscribers how to use AI to secure commissions — advertising slots at outlets like Dispatch as paying "around £400 per feature" and charging members 1,200 Kenyan shillings (~£6.90/$10) per month. [1][2][5]
  • Sure Media Agency recruits via TikTok videos, reaching young Kenyan writers who use the platform to find international commissions without relying on traditional job boards or industry contacts; one profiled subscriber, 25-year-old Adrian Waweru, is an IT officer for Kenya's Ministry of Education who turned to the service after seeing an ad while scrolling his phone. [1][2][5]
  • At the AJEN Roundtable in Mauritius, questions were raised about equity within Kenya's journalism AI ecosystem — specifically who controls access to AI opportunities and who benefits from them in Kenyan newsrooms. [3][4][6]
Sure Media AgencyDispatchAdrian WaweruAJEN RoundtableNairobiMauritiusKenya's Ministry of EducationTikTokAI-generated freelance pitchesKenyan media platformAI in journalismgig economy journalismAI equity in newsroomsinternational news commissionsAfrican freelance writers
Sources
  1. [1]pressgazette — A journalist at Dispatch has investigated the ecosystem that has led to writers in Kenya inundating commissioning editors in the UK and US wtwitter
  2. [2]julianrademeyer — Inside Kenya's AI journalism factory. Editors are being inundated with AI-generated pitches. But who's behind them? https://t.co/JWfZtBs3Sntwitter
  3. [3]AKU_GSMC — Who gets access to AI opportunities in Kenya’s journalism ecosystem and who controls the pathways? At the AJEN Roundtable in Mauritius, Proftwitter
  4. [4]AKU_GSMC — Who gets access to AI opportunities in Kenya’s journalism ecosystem and who controls the pathways? At the AJEN Roundtable in Mauritius, Proftwitter
  5. [5]julianrademeyer — Inside Kenya's AI journalism factory. Editors are being inundated with AI-generated pitches. But who's behind them? https://t.co/JWfZtBs3Sntwitter
  6. [6]AKU_GSMC — Who gets access to AI opportunities in Kenya’s journalism ecosystem and who controls the pathways? At the AJEN Roundtable in Mauritius, Proftwitter
11.41

Lettertrace tool tracks brand mentions across AI systems for enterprise

  • Lettertrace, launched by The Letter Company and going viral on Product Hunt, is a free, open-source tool that measures how often AI systems — specifically ChatGPT, Claude, and Gemini — mention a given brand, positioning AI visibility as a trackable business metric. [1][4][5]
  • The tool directly undercuts an existing pricing model in the AI brand-monitoring space, where competitors charge $250/month for equivalent brand-visibility data from a single AI system like ChatGPT. [2][3]
  • HuggingPapers highlighted LLMRouter, a unified open-source infrastructure for LLM routing that covers 16+ methods spanning single-turn, multi-turn, and personalized settings. [6]
  • OpenRouter, which can observe which AI models developers actually select across every provider, is hiring a data-focused role at a salary range of $180,000–$240,000 to mine that model-selection data. [7]
LettertraceThe Letter CompanyChatGPTClaudeGeminiLLMRouterHuggingPapersOpenRouterAI brand visibility monitoringopen-source LLM toolsAI model routing infrastructurebrand mention trackingAI market competitiondeveloper model selection dataLLM routing methods
Sources
  1. [1]mathewpregasen — Today is a really exciting day for The Letter Company. Lettertrace is going viral on Product Hunt! A lot of people have asked why we made Le★twitter
  2. [2]Suryanshti777 — $250/mo to find out what ChatGPT says about your own company. Wild that anyone's paying that Lettertrace just torched that pricing model. Op★twitter
  3. [3]shedntcare_ — wait... how often does ChatGPT actually mention your company? Lettertrace lets you measure that across Claude, ChatGPT and Gemini and it's otwitter
  4. [4]heyshrutimishra — AI search is becoming a real discovery channel for products Found Lettertrace: free, open-source, and it actually tells you what ChatGPT, Cltwitter
  5. [5]iam_chonchol — AI visibility is becoming a metric you probably want to know before you need it People already ask ChatGPT, Claude and Gemini which tools totwitter
  6. [6]HuggingPapers — LLMRouter A unified open-source infrastructure for LLM routing, covering 16+ methods across single-turn, multi-turn, and personalized settintwitter
  7. [7]hotfixjobs — OpenRouter can see which models developers actually pick, across every provider. They're hiring someone to mine that data. $180k to $240k, ftwitter
11.42

Unitree Robotics IPO oversubscribed 8,000x at $9B valuation

  • Unitree Robotics priced its IPO on the Shanghai exchange, raising approximately $900M for a roughly 10% stake, implying a valuation of more than $9 billion; the offering became the first humanoid robot company to go public on a mainland Chinese exchange. [4][7]
  • The Unitree IPO was reportedly oversubscribed approximately 8,000 times by retail investors, making it one of the most oversubscribed offerings on record and reflecting intense investor appetite for humanoid robotics following the AI investment cycle. [3][5][7]
  • Unitree stated that cumulative production of its bipedal humanoid robots reached approximately 18,000 units by the end of July, spanning their full bipedal product lineup. [6]
  • Commentators noted that Unitree sells humanoid robots at a price point below that of a maxed-out iPhone, framing the offering as a signal that humanoid robotics pricing has moved closer to consumer accessibility rather than remaining a distant-future technology. [1]
  • Approximately 90 companies are now reported to be pursuing physical AI and humanoid robotics, with Unitree cited as the first to reach public markets but not the only entrant developing such technology. [2]
Unitree RoboticsShanghai exchangehumanoid robotics IPObipedal robot manufacturingphysical AI investmentmainland China stock marketrobotics valuationconsumer-accessible robotics pricinghumanoid robot oversubscriptionrobotics industry competition
Sources
  1. [1]panter_0x — UNITREE SELLS A HUMANOID CHEAPER THAN A MAXED OUT IPHONE. THIS ISN'T A SALE. The market still files humanoids under "distant future technolotwitter
  2. [2]zerokn0wledge_ — The humanoid robot race is accelerating! $UNITREE is the first to go public, but not the only one cooking. 90 companies are now chasing phystwitter
  3. [3]aistasiia — FROM THE AI BUBBLE TO THE ROBOT BUBBLE Investors are pouring into humanoids. Unitree’s IPO was oversubscribed 8,000×. $9B valuation. 5,500+ twitter
  4. [4]CSISFreeman — China's leading humanoid robot maker, Unitree, has priced its IPO at a valuation of more than $9 billion, becoming the country's first mainltwitter
  5. [5]BoomingBulls — China’s IPO market is showing an unbelievable level of retail investor frenzy! 🤯 Chinese humanoid robot maker Unitree reportedly saw its IPOtwitter
  6. [6]korthosxyz — Unitree has stated cumulative production of its bipedal humanoid robots reached ~18,000 units by the end of July. This figure spans their bitwitter
  7. [7]gulVasikova — China is sending a huge signal on physical AI and robotics. 🤖🇨🇳 Unitree Robotics’ ~$900M Shanghai IPO attracted more than 8,000× retail overtwitter
11.43

X Open-Sources Rebuilt "For You" Algorithm With Transparency Tool

  • X released its rebuilt "For You" feed ranking algorithm source code on GitHub under the Apache v2 license, described as a complete replacement of the old rules-based system with a new ML-based ranking model — the legacy system is characterized as entirely gone. [5][8]
  • Alongside the code release, X launched an "Under the Hood" visibility tool that lets users see why specific posts are being ranked or suppressed in their feed, addressing longstanding creator complaints about so-called "shadowbanning." [2][3]
  • The released repository contains approximately 363,000 lines of code; analysts reviewing it found that the system rewards predicted viewer actions rather than raw engagement counts, meaning X's ranking is not primarily optimizing for likes or reposts as standalone metrics. [4][7]
  • Reviewers of the scoring weights found that replies carry the most weight of any engagement type in the ranking system. [6]
  • The code also contains a hardcoded mechanic that drops one account with fewer than 1,000 followers into slot 15 of a user's feed, functioning as a built-in lottery for smaller accounts. [1]
GitHubApache v2feed ranking algorithmopen source transparencyML-based ranking modelshadowbanning visibility toolengagement weightingcreator discoverabilityalgorithmic accountabilityreply signal prioritysmall account promotion
Sources
  1. [1]gemchange_ltd — X open-sourced the For You algorithm. There's a hardcoded lottery that drops one sub-1k account into slot 15 of your feed. There's a number twitter
  2. [2]SlayHouseLive — 🚨 X JUST GAVE YOU A WAY TO SEE IF THE ALGORITHM HAS BEEN BURYING YOUR POSTS After YEARS of creators screaming “SHADOWBAN!” every time their twitter
  3. [3]ExtensiveGrowth — $SPCX X open-sources For You ranking code + launches “Under the Hood” visibility tool X took another step on algorithm transparency: a much twitter
  4. [4]plbiojout — X just open-sourced the closest thing we’ve ever had to the actual “For You” algorithm. Me and Claude read the 363,000 lines they released ttwitter
  5. [5]cigale_ai — 🚨 X just open-sourced the completely rebuilt "For You" algorithm. The old rules are dead. The legacy system is gone, replaced entirely by a twitter
  6. [6]rehmanrindd — 🚨 X just open sourced the For You algorithm. The scoring weights are in the repo. TL;DR: > Replies carry the most weight of any engagement, twitter
  7. [7]hvxcrypto — You need to know this regarding X 'For you' algorithm: X rewards predicted viewer actions, not raw engagement counts. X is not primarily asktwitter
  8. [8]ShowsAli — X released its 'For You' algorithm source code on GitHub this Thursday under the Apache v2 license. The update includes a transparency tool twitter
11.44

Lockheed Martin Demos 5G AI Drone Detection Service With Verizon, NVIDIA

  • Lockheed Martin, alongside Verizon, NVIDIA, Keysight Technologies, startup ODC, and Lockheed's own Astris AI subsidiary, conducted a live demonstration in the Miami area of their NetSense Airspace Awareness-as-a-Service system, which detects, tracks, and monitors unmanned aircraft systems (UAS) using existing 5G networks without requiring any changes to radio deployments or dedicated radar hardware. [1][2][4]
  • The NetSense system works by reading RF disturbances in Verizon's existing 5G spectrum via ODC's AI-native Radio Access Network (AI-RAN) software, NVIDIA's AI Aerial platform (which analyzes signal disturbances and feeds results into NetSense algorithms), and Keysight's RF simulation platform, with Lockheed's warning and tracking technology then predicting drone flightpaths and automatically issuing real-time alerts. [2][4]
  • Lockheed Martin plans to offer NetSense as a subscription service integrated with customers' security operations; a prototype was unveiled in March, pilot deployments are expected in the latter half of this year and early next year, and general availability is planned for 2027. [1][3]
  • Lockheed Martin identified broad commercial use cases for NetSense, including critical infrastructure protection, stadiums, power plants, schools, hospitals, and airports, as well as potential future integration with 6G Integrated Sensing and Communication (ISAC) technology, which would transform mobile networks from data-delivery systems into spatial-sensing platforms. [1][2][4]
Lockheed MartinVerizonNVIDIAKeysight TechnologiesODCAstris AINetSenseAI AerialAI-RANdrone detection and tracking5G network airspace awarenessunmanned aircraft systems monitoringRF disturbance sensingAI-native radio access networkairspace-as-a-servicecritical infrastructure protection6G integrated sensing and communicationcounter-UAS technologysubscription security service
Sources
  1. [1]Light_Reading — Lockheed Martin plans to offer a 5G drone detection service next year using Verizon's network and tech from Nvidia, ODC, Keysight and subsidtwitter
  2. [2]TelecomTV — US aerospace giant @LockheedMartin has partnered with @Verizon, @nvidia and others to demonstrate how existing #5G networks can power AI-enatwitter
  3. [3]shehzadyounis — Lockheed Martin preps drone detection service using Verizon 5G Lockheed Martin plans to offer a 5G drone detection service next year using Vtwitter
  4. [4]everythingRFcom — @LockheedMartin and Partners Demonstrate AI-Powered 5G System for Drone Detection Read more - https://t.co/e7cDpVEIB4 #ai #5G #partnership #twitter
11.45

Sakana AI updates Sakana Chat with Fugu, Namazu models and vibe-coding

  • Sakana AI updated Sakana Chat to integrate two next-generation models: Sakana Fugu, an orchestrator model that dynamically combines multiple AI models to handle complex, multi-step tasks and claims performance rivaling frontier models; and an updated Sakana Namazu, which switched its base from DeepSeek-V3.1-Terminus to Moonshot AI's open model Kimi K2.6, improving Japanese-language response quality and autonomous task execution. The update is available free with no login required. [1][3][4][5]
  • Sakana Fugu, previously accessible only via API, is now usable directly in the browser — including on mobile — through the Sakana Chat interface, though selecting it requires email registration. [3][6]
  • The updated Sakana Chat adds code execution capabilities: models can run Python in a sandboxed environment to perform calculations, data processing, and file creation, with outputs including graphs, HTML previews, Word documents, slides, and simple games displayed in a side panel for immediate preview or download. [3][4][5]
  • Sakana Chat also gains support for file attachments including Word, Excel, and PDF formats, and supports vibe-coding — generating interactive apps and games from natural-language descriptions — with demonstrated use cases including an interactive math game for kids and a fully cited market research report. [2][3][5][7]
Sakana AISakana ChatSakana FuguSakana NamazuDeepSeek-V3.1-TerminusMoonshot AIKimi K2.6AI model orchestrationmulti-step task handlingJapanese-language AIcode execution sandboxvibe-codingfile attachment supportbrowser-based AI interfacefrontier model performancenatural-language app generationinteractive content generation
Sources
  1. [1]SakanaAILabs — Sakana Chat just got a big upgrade. No login required, free to use: https://t.co/lyIrGJXwqo Powered by Fugu and Namazu, our Japanese LLM. Witwitter
  2. [2]hardmaru — Here is a quick 60-second summary of vibe-coding in Sakana Chat: • An interactive math game for kids • A fully cited market research report twitter
  3. [3]itm_aiplus — 「Sakana Fugu」が無料で試せるように チャットAIをアップデート、Sakana AI https://t.co/HreUYyDjIRtwitter
  4. [4]SakanaAILabs — Sakana Chatの大規模アップデートを公開しました。 ログイン不要・無料でご利用いただけます：https://t.co/stLdPSzzNF Fuguと最新の日本語LLM「Namazu」を搭載し、待望のコード実行機能を追加しました。日本語で作りたいものを説明するだけで、Wetwitter
  5. [5]madonomori — Sakana AIの「Sakana Chat」で「Sakana Fugu」を利用可能に ～「Sakana Namazu」も更新／Word文書やゲームなどの成果物作成やファイル添付も… https://t.co/Ks9Rke2Om6 https://t.co/WrKzqUGmKxtwitter
  6. [6]ChiekoTazuke — Sakana Fuguというモデルオーケストレーターですが、いままでAPIのみの提供でしたが、ブラウザからチャット上で使えるようになりました。スマホからでもバイブコーディングできる🐟！twitter
  7. [7]hardmaru — Here is a quick 60-second summary of vibe-coding in Sakana Chat: • An interactive math game for kids • A fully cited market research report twitter
11.46

Google Cloud Signs Five-Year AI and Cloud Partnership with Ryanair

  • Ryanair and Google Cloud announced a five-year partnership on August 12, 2026, under which Ryanair will deploy Google Workspace and Google Cloud services to 35,000 employees across its network, with the stated goal of supporting growth to 300 million passengers per year by 2034. [2][1]
  • The deal includes Gemini Enterprise — Google Cloud's agentic AI platform designed to connect organisational data, automate workflows, and create custom AI agents — specifically to automate decision-making, optimise flight crew logistics, and support overall corporate productivity. [2][4]
  • Beyond Gemini, the partnership also targets deployment of two DeepMind research models: AlphaEvolve (combinatorial optimization) and WeatherNext (weather forecasting), directed at fleet operations, maintenance scheduling, and weather-driven disruption management — described as the more structurally significant elements of the deal. [3]
  • Ryanair is retaining its existing Amazon Web Services infrastructure while adding Google Cloud as a second provider, making this a deliberate dual-cloud/multi-cloud redundancy strategy rather than a full platform migration. [3][2]
RyanairGoogle CloudDeepMindGemini EnterpriseAlphaEvolveWeatherNextGoogle WorkspaceAmazon Web Servicesairline AI partnershipmulti-cloud strategyagentic AI platformflight crew logistics optimizationcombinatorial optimizationweather forecasting for aviationworkflow automationfleet operations and maintenancecorporate productivitycloud infrastructure redundancy
Sources
  1. [1]NewsFromGoogle — Ready for takeoff: @GoogleCloud and @Ryanair are launching a new 5-year partnership. ✈️ Europe's largest airline will use Gemini Enterprise twitter
  2. [2]AkwyZ — Ryanair 🤝 Google Cloud A new 5-year partnership is bringing Google Workspace & Gemini Enterprise to 35k Ryanair employees! ✈️ They're ustwitter
  3. [3]fourweekmba — Ryanair and Google Cloud's Five-Year Deal Shows Where Enterprise AI Actually Earns Its Keep https://t.co/Ayiuqsc3vs #AI #Google https://t.cotwitter
  4. [4]metaintro — LABOR DESK: Ryanair is putting Google's Gemini Enterprise to work on flight crew logistics and rolling Google Workspace out to 35,000 staff twitter
11.47

Tencent Q2 2026 Revenue Rises 11% on AI Advertising and Gaming

  • Tencent reported Q2 2026 revenue of 204.8 billion yuan ($30.4 billion), an 11% year-over-year increase that beat analyst consensus estimates, driven by AI-enhanced WeChat advertising and a rebound in gaming sales. [1][2][5][7][9]
  • China gaming sales jumped 17% and AI-driven advertising on WeChat were the two primary contributors to the revenue beat, according to the reported results. [5][6][8]
  • Despite the strong top-line performance, heavy AI investment spending is pressuring Tencent's bottom line, creating a tension between robust revenue growth and rising costs associated with AI infrastructure. [4][9]
  • The quarter was described as providing tangible evidence that Tencent's AI investment is beginning to generate direct revenue benefits, particularly in advertising. [3]
TencentWeChatTencent quarterly earningsAI-driven advertising revenueChina gaming sales growthAI infrastructure investmentrevenue beat analyst estimatesWeChat monetizationAI return on investment
Sources
  1. [1]techshotsapp — AI Wins the Ad War, But Tencent’s Bottom Line Takes the Hit Tencent reported an 11% rise in Q2 revenue to 204.8 billion yuan ($30.4 billion)★twitter
  2. [2]Cbk19881 — $TCEHY Q2 2026 Results ​Tencent Holdings reported second quarter 2026 financial results today, with top line growth beating consensus while twitter
  3. [3]EmmanuelInvest — 🤖 AI — THE BIG STORY This quarter provides tangible evidence that Tencent's AI investment is beginning to generate revenue benefits, particutwitter
  4. [4]VicentZacky — TENCENT IS SPENDING BILLIONS TO WIN THE AI WAR — AND THE NUMBERS ARE SCREAMING Tencent just delivered a quarter that looks strong on the surtwitter
  5. [5]wlst_news — Tencent’s Q2 revenue rose 11% to 204.8 billion yuan, beating expectations as China gaming sales jumped 17% and AI helped boost advertising, twitter
  6. [6]BreakingEquity — Tencent ($TCEHY) posted a Q2 revenue beat, with results helped by stronger game sales and better performance in AI-driven advertising. Anothtwitter
  7. [7]UserNai002 — 🇨🇳 Tencent Q2 results beat estimates on robust AI-enhanced advertising Tencent, China's most valuable company, saw its second-quarter sales twitter
  8. [8]AfricaisHOME2 — Tencent topped sales estimates after a surge in WeChat advertising and a rebound in its core gaming business. The WeChat owner reported anottwitter
  9. [9]Sysyx63 — 🇨🇳 Tencent Q2 2026: strong operating growth, but AI investments are becoming increasingly expensive. Tencent reported revenue of RMB 204.8 btwitter
11.48

Cerebras Q2 2026 Core Revenue Doubles, Cloud Revenue Nearly Quadruples

  • Cerebras Systems reported Q2 2026 core revenue of $209.9 million, up 103% year-over-year, while GAAP cloud and other services revenue reached a record $126.0 million, up 281% year-over-year. [3][6][10]
  • Cerebras' fast inference cloud business nearly quadrupled year-over-year, representing the primary driver of growth in the quarter. [1][2]
  • Total GAAP revenue came in at $180.11 million versus analyst consensus estimates of $193.55 million — a 9% miss — despite beating the company's own revenue guidance by 8%; the stock fell 11.5% following the report. [4][5]
  • Cerebras CEO Andrew Feldman stated the company beat guidance on every core business metric — revenue, gross profit, and adjusted EBITDA — and raised its full-year 2026 guidance following the results. [2][7][8]
  • The company also reported an adjusted net loss per share of $(0.28) and cited a $25 billion backlog as a key forward-looking indicator alongside its IPO proceeds. [7][9]
Cerebras SystemsAndrew FeldmanAI inference cloud revenuequarterly earnings resultsyear-over-year growthrevenue guidance beatfast inference cloudadjusted EBITDAbacklog forecastIPO proceedsstock reactionanalyst consensus miss
Sources
  1. [1]cerebras — Cerebras today announced financial results for Q2 2026, with our fast inference cloud business nearly quadrupling year over year. • GAAP clo★twitter
  2. [2]andrewdfeldman — A Good Q2 for @cerebras. Cloud revenue nearly quadrupled. Core revenue more than doubled. Beat guidance on every core business metric - revetwitter
  3. [3]AlphaSenseInc — $CBRS Earnings: - Record GAAP cloud and other services revenue of $126.0 million, up 281% year-over-year; Record core cloud and other servictwitter
  4. [4]Schulz_Research — $CBRS beat its own revenue guidance by 8% tonight. It also missed consensus by 9%. Both are true. The stock is down 11.5%. The gap between ttwitter
  5. [5]MikeLongTerm — BREAKING $CBRS Q2 2026 Earning Result🚨🚨🚨 Revenue: $180.11M vs $193.55M est EPS: -$2.98 vs -$167 est “This was an outstanding quarter for Certwitter
  6. [6]athuinvests — $CBRS Q2 Earnings Notes: > Andrew Feldman, CEO - "This was an outstanding quarter for Cerebras. Core revenue more than doubled to $210 millitwitter
  7. [7]allday_stocks — $CBRS 𝐂𝐥𝐨𝐮𝐝 𝐑𝐞𝐯𝐞𝐧𝐮 𝐍𝐞𝐚𝐫𝐥𝐲 𝐐𝐮𝐚𝐝𝐫𝐮𝐩𝐥𝐞𝐬 𝐚𝐬 𝐂𝐞𝐫𝐞𝐛𝐫𝐚𝐬 𝐃𝐨𝐮𝐛𝐥𝐞𝐬 𝐂𝐨𝐫𝐞 𝐑𝐞𝐯𝐞𝐧𝐮𝐞, 𝐑𝐚𝐢𝐬𝐞𝐬 𝐅𝐘𝟐𝟔 𝐆𝐮𝐢𝐝𝐚𝐧𝐜𝐞 ⠀ 📊 𝐑𝐞𝐬𝐮𝐥𝐭𝐬 • Adj. net loss per share: $(0.28) •twitter
  8. [8]YeboahWalee — $CBRS 𝐂𝐥𝐨𝐮𝐝 𝐑𝐞𝐯𝐞𝐧𝐮 𝐍𝐞𝐚𝐫𝐥𝐲 𝐐𝐮𝐚𝐝𝐫𝐮𝐩𝐥𝐞𝐬 𝐚𝐬 𝐂𝐞𝐫𝐞𝐛𝐫𝐚𝐬 𝐃𝐨𝐮𝐛𝐥𝐞𝐬 𝐂𝐨𝐫𝐞 𝐑𝐞𝐯𝐞𝐧𝐮𝐞, 𝐑𝐚𝐢𝐬𝐞𝐬 𝐅𝐘𝟐𝟔 𝐆𝐮𝐢𝐝𝐚𝐧𝐜𝐞 ⠀ 📊 𝐑𝐞𝐬𝐮𝐥𝐭𝐬 • Adj. net loss per share: $(0.28) •twitter
  9. [9]EmmanuelInvest — 🚨 $CBRS (Cerebras Systems) Q2 2026 Earnings Explosive cloud growth… IPO war chest + $25B backlog are the real story 👀 📊 KEY METRICS (Q2 2026twitter
  10. [10]alexanderbittan — @cerebras just delivered a much stronger quarter than the headline GAAP numbers suggest. • Core revenue: $209.9M, up 103% YoY and above its twitter
11.49

Kalshi becomes first prediction market to stream real-time order book data over dedicated fiber network

  • Kalshi claims to be the first prediction market to distribute real-time order book data over a dedicated fiber network, streaming its live order book through DoubleZero Edge — the same type of low-latency infrastructure used by major exchanges such as NYSE and CME. [1][2]
  • The DoubleZero x Kalshi integration makes institutional-grade, full-depth-of-book market data accessible to prediction market traders, a capability previously reserved for major financial institutions. [3][4][5]
KalshiDoubleZero EdgeNYSECMEprediction marketsreal-time order book datalow-latency infrastructurefiber networkinstitutional-grade market datafull-depth-of-bookfinancial exchanges
Sources
  1. [1]CoinMarketCap — LATEST: ⚡ Kalshi says it has become the first prediction market to distribute real-time order book data over a dedicated fiber network, givi★twitter
  2. [2]ITzINSANE_NFTs — The same data infrastructure NYSE and CME have been running on for decades just opened up to anyone trading prediction markets. DoubleZero xtwitter
  3. [3]santotheplug1_ — DoubleZero x Kalshi 🥶 Bringing a piece of market infrastructure that was usually reserved for major financial institutions into prediction mtwitter
  4. [4]Alexvx_nft — THIS USED TO BE A CLOSED ROOM multi-monitor desk dedicated fiber full depth of book streaming in real time that stack lived at NYSE and CME twitter
  5. [5]laki_0x — this is bigger than just faster data DoubleZero x Kalshi is giving prediction market traders access to infrastructure that was basically restwitter
11.50

Qodo Launches Free AI Code Review Academy for Enterprise Teams

  • Qodo launched the AI Code Review Academy, a free, structured learning hub written by developers and ML engineers with 10+ years of experience, targeting engineering teams evaluating, implementing, or scaling AI code review; every chapter originated from a real question posed by a real engineering team. [1][5]
  • The Academy's content frames AI code review as a dedicated verification discipline — distinct from code generation tools — covering what to evaluate, how it fits within the SDLC, how to run a meaningful proof of concept, and how to define KPIs and benchmarks. [1][2]
  • The Academy explicitly addresses failure modes specific to AI-generated code — including logic errors, duplicated logic, breaking changes, security vulnerabilities, and standards drift — arguing that a dedicated verification layer between AI code generation and production is "non-negotiable at enterprise scale." [1][3][4]
  • One chapter tackles tool selection in a crowded market where vendors claim identical outcomes, outlining five evaluation criteria to differentiate tools; the content was written by engineers and ML researchers building AI code governance solutions. [1][5]
QodoAI Code Review AcademySDLCAI code reviewcode review automationAI-generated code verificationenterprise software developmentproof of concept evaluationKPIs and benchmarkscode governancesecurity vulnerabilities in AI codesoftware development lifecycletool selection criteria
Sources
  1. [1]QodoAI — 🎓 The Qodo AI Code Review Academy is live. Learn how to evaluate tools, understand benchmarks, define KPIs, run a meaningful POC, build the twitter
  2. [2]nnennahacks — Qodo’s AI Code Review Academy is live. A practical guide to AI code review a lot of team effort went into! I’m particularly excited about thtwitter
  3. [3]nnennahacks — Qodo’s AI Code Review Academy is live. A practical guide to AI code review a lot of team effort went into! I’m particularly excited about thtwitter
  4. [4]nnennahacks — Qodo’s AI Code Review Academy is live. A practical guide to AI code review a lot of team effort went into! I’m particularly excited about thtwitter
  5. [5]QodoAI — 🎓 The Qodo AI Code Review Academy is live. Learn how to evaluate tools, understand benchmarks, define KPIs, run a meaningful POC, build the twitter
11.51

YC-backed Prodigy Research claims AI quant exceeds 100% live trading returns

  • Prodigy Research (YC S26) launched publicly as a self-described "frontier AI trading research lab," claiming their AI quant achieved over 100% returns in live trading over approximately two months using a delta-neutral strategy, with no losing weeks during that period. The company further claims their model outperforms a top-10% trader at Jane Street in internal testing. [3][2][4][6]
  • Skepticism from industry insiders followed the announcement, with a quant fund founder and former Bridgewater quant raising the question of why, if the returns are genuine, the company would go through YC and raise venture capital rather than simply trade its own capital. [7]
  • A Series 57-licensed securities trader at Jane Street publicly questioned the claims, adding to broader industry pushback. [1][5]
  • Critics also noted that major HFT funds on Wall Street have long been training their own models on vast proprietary trading datasets, questioning the novelty of Prodigy Research's approach. [5]
Prodigy ResearchY CombinatorJane StreetBridgewaterAI quantitative tradingalgorithmic trading claimsdelta-neutral strategyfrontier AI trading researchventure capital skepticismHFT competitionlive trading performanceproprietary trading models
Sources
  1. [1]DeepDishEnjoyer — "Series 57-licensed securities trader at Jane Street" https://t.co/VlpZ4B0aSe — quoting @michaelwangelo: Today, we’re launching Prodigy Rese★twitter
  2. [2]0xLouisT — - "our AI quant outperforms a top 10% trader at Jane Street" - "we've already achieved more than 100% returns in live trading over our YC batwitter
  3. [3]michaelwangelo — Today, we’re launching Prodigy Research (YC S26), the frontier AI trading research lab. We’re training the world’s best foundation model fortwitter
  4. [4]Anitahityou — 🚨 YC 孵化的量化 AI 宣布打败 Jane Street YC S26 的 Prodigy Research 声称，他们的 AI quant 用 Delta-neutral 策略实盘两个月做到 +108%，期间没有一个亏损周，内部测试甚至超过 Jane Street 前 10twitter
  5. [5]RehanNek — ummm.....isn't like almost every HFT fund on wall street training their own model with a gazillion proprietary trading data points? — quotintwitter
  6. [6]launcharchive — Prodigy Research's AI quant delivers over 100% returns, outperforming elite traders in quantitative finance. @michaelwangelo reached #2 on Ltwitter
  7. [7]shellvish — Thoughts from a quant fund founder + former Bridgewater quant: Many people have asked the obvious question: why go through YC and raise VC ctwitter