US Commerce withdraws June export controls on Claude Fable 5, restoring global enterprise access to Anthropic's most powerful model with conditions. ⚖️
The AI Daily
Daily · July 2, 2026
Happy Thursday. Two big moves dominated the day: Meta's surprise plan to sell excess AI infrastructure capacity to third parties sent its stock up 9%, signaling that hyperscaler AI spend can now become a revenue line, not just a cost center. Meanwhile, the US lifted emergency export controls on Anthropic's Claude Fable 5, restoring global enterprise access to its most powerful model — but with strings attached. Together, the stories hint at a maturing AI stack where compute is a commodity and geopolitics shape who gets to use frontier models.
🧠 Foundation models
go deeper →The US government's reversal on Claude Fable 5 export controls is the headline story in foundation models today — Anthropic's most capable model is back on the global market, but the episode underscores how frontier AI has become entangled with national security policy.
Anthropic is bringing back Claude Fable 5 globally after US lifts export control order — where can enterprises access it?
🏗️ Infrastructure
go deeper →Meta's pivot to monetize its AI compute surplus is the infrastructure story of the day, while Together AI's $800M raise and a flurry of global data center commitments — from South Korea's $919B mega-project to a Valar-Nvidia nuclear data center partnership — underscore that the build-out race is far from over.
Meta pops 9% as company makes cloud push to sell excess AI compute power capacity
Meta will sell unused AI infrastructure capacity to third parties, turning aggressive capex into a revenue stream and easing investor concerns. 💰
Together AI raises $800M to grow its AI-optimized public cloud
Together AI's massive Series C, led by Aramco Ventures with Nvidia participating, cements the open-source AI cloud segment as a serious infrastructure tier. 💰
South Korea announces $919bn investment into three "mega projects," plans to build 18.4GW worth of data centers by 2035
South Korea's sovereign AI infrastructure bet — 18.4GW of data centers by 2035 — is among the most ambitious national compute programs announced to date. ⚖️
Valar nuclear startup partners with Nvidia on data center aiming to conserve water
Nuclear startup Valar pairs with Nvidia to power a water-conserving AI data center, spotlighting new energy-infrastructure partnerships for AI compute.
Cloud Capital and Realty Income partner for $6bn data center joint venture fund
⚡ A $6B real-estate-backed data center JV acquiring Virginia sites signals institutional capital is firmly embedding itself in AI infrastructure. 💰
AWS launches Secret Cloud for Industry
AWS's classified cloud offering for US defense contractors expands the secure AI compute ecosystem for national security workloads. 📱
💰 Funding & deals
go deeper →The funding circuit is active across the stack: Together AI's $800M Series C is the headline raise, while Raja Koduri's Oxmiq Labs (custom AI silicon) and Venice.ai (private uncensored AI) reflect continued investor appetite from infrastructure to application layer.
Venice raises $65M at $1B valuation for private, uncensored AI
Venice.ai's unicorn raise signals investor demand for privacy-first AI alternatives to mainstream chatbots like ChatGPT. 📱
Raja Koduri's Oxmiq Labs raises $35M to lower the design cost of custom AI silicon
Ex-Intel/AMD chip architect Raja Koduri's Oxmiq raises $35M to democratize custom AI silicon design, aiming to be the Arm of AI chips. 🏗️
EXCLUSIVE: SoftBank renews talks for $10 billion loan against OpenAI stake, adds concessions, sources say
SoftBank's renewed $10B loan talks against its OpenAI stake, with added concessions, keeps one of AI's biggest financial bets in motion.
🔧 Middleware & platforms
go deeper →Cloudflare is forcing a reckoning on AI web crawlers — giving companies until September to separate search from training bots or face publisher blocks — while AWS upgrades its OpenSearch log analytics engine at no extra cost, reflecting the quiet but important battle over data rights and platform economics.
Cloudflare's new policy pushes AI companies to pay for publishers' content
Cloudflare's September deadline to separate AI training crawlers from search crawlers could reshape how AI companies access the open web. ⚖️
AWS turbocharges log analytics in Amazon OpenSearch Service at no extra cost
AWS's new purpose-built log analytics engine cuts OpenSearch storage costs ~70%, making AI-adjacent observability significantly cheaper at scale.
The Control Gap: Enterprise AI organizations have an ownership problem, not a technology problem
New research finds most enterprises govern sprawling AI portfolios manually, with no reliable way to detect model drift — a governance gap that middleware vendors should fill.
📱 Application solutions
go deeper →AI is moving deeper into physical and transactional workflows: Verkada integrates Nvidia AI across 2.4 million security devices, Square enables ChatGPT and Claude to place restaurant orders directly, and Palantir's CEO publicly attacks the token-cost economics of OpenAI and Anthropic as unsustainable for enterprise buyers.
Verkada takes Nvidia investment to expand its physical AI platform
⚡ Nvidia's investment in Verkada accelerates AI inference across 2.4 million connected physical security devices, a major edge deployment signal. 💰
Restaurants can now accept orders placed directly from ChatGPT and Claude thanks to Square's new, low-fee, no setup integration
⚡ Square's ChatGPT and Claude plugins let consumers order food through AI chat, putting agentic commerce into live production for restaurants. 🔧
Palantir's Karp bashes OpenAI, Anthropic token model: 'Something has gone completely wrong'
Palantir's CEO argues skyrocketing token costs are pushing enterprises toward open-weight models, flagging a structural pricing tension in the AI market. 🧠
Telangana police use AI-powered drones with facial recognition for Hyderabad street patrolling
⚡ Hyderabad police deploying AI facial-recognition drones for routine street surveillance marks a significant — and contested — public-sector AI rollout. 🇮🇳 ⚖️
⚖️ Policy & legal
Two US government actions dominated policy today: the FTC warned that AI bias safeguards could paradoxically violate consumer protection law, while the Commerce Department lifted export controls on Anthropic's Claude Fable 5 — but only after Anthropic added new security measures to appease the Trump administration.
Anthropic Added a New Security Measure to Get Back Into the Trump Administration's Good Graces
Anthropic's compliance concession to lift Claude Fable 5 export controls reveals how frontier AI companies must navigate US national security politics. 🧠
US FTC says AI bias safeguards may run afoul of consumer law
The FTC's warning that AI fairness guardrails could breach consumer law creates a regulatory paradox for companies trying to build responsible AI.
Inside China's grey market for banned US AI models: Transfer station economy explained
⚡ A thriving Chinese API-proxy grey market for Claude and other banned US models shows that export controls create black markets, not hard barriers. 🧠
🇮🇳 India lens
India's AI story today spans surveillance and infrastructure: Telangana police are deploying AI facial-recognition drones for routine Hyderabad street patrols — a live production rollout with significant civil liberties implications — while Tata Communications announces India-Singapore subsea cable investments that will underpin the country's growing data economy.
Tata Communications announces India-Singapore subsea cable investments
Tata Communications' Mumbai-Singapore and Chennai-Singapore subsea cables will materially improve India's AI data pipeline capacity. 🏗️
Everforth bets on India's GCC boom, AI demand to drive growth
US tech services firm Everforth sees India's maturing GCC ecosystem and AI-led engineering demand as its primary growth engine post-rebrand. 📱
📖 Beyond the headlines
The Control Gap: Enterprise AI organizations have an ownership problem, not a technology problem — and most are governing it by hand — While the industry debates foundation model capabilities and infrastructure scale, this piece exposes the unglamorous reality inside enterprise AI programs: contested platform ownership, manual governance, and no reliable drift detection. For strategists evaluating where the next wave of B2B AI software value will be created, this governance gap is the clearest unmet need in the stack. Read →