TechWithAdit (TWA)

The 18-day AI export war is over. Trump lifted all restrictions on Anthropic's Fable 5 and Mythos 5 late Tuesday night, and Anthropic began restoring global access this morning — July 1 — to all users. The ban that locked out international developers, triggered legal action, and forced the Five Eyes to issue their most urgent AI warning in history is done. What Anthropic agreed to in return is the more important story.

While that resolves, the internet's original architect is stepping down. Vint Cerf, 83 — who co-designed the TCP/IP protocols that built the internet we know — is retiring from Google next week. And a Czech startup founded by three ex-DeepMind researchers who once built a poker AI has just hit a $500M valuation by applying the same reinforcement learning to the S&P 500, with a perfect record of zero negative months since inception.

Vinton Cerf, 83, will step down as Google's Chief Internet Evangelist next week, closing one of the most consequential careers in technology history. Speaking via video at the Open Frontier conference hosted by the Laude Institute, Cerf was recognised by UC Berkeley professor Dave Patterson — co-developer of the RISC processor architecture — with an announcement that drew a standing ovation. Cerf and collaborator Robert Kahn co-designed TCP/IP in the 1970s, the foundational set of rules that lets different computer networks communicate with each other, and the reason the internet you use today works the way it does. He joined Google in 2005 and has spent 20+ years working on internet accessibility, policy, and global infrastructure — while also collecting the Presidential Medal of Freedom, the Turing Award, and 29 honorary degrees.

What made his farewell remarks striking was their forward-looking clarity. In a conference focused on the centralisation of AI in a handful of labs — a sharp contrast to the decentralised internet Cerf built — he offered a prediction: AI agents will eventually force the industry back toward open standards. "The agentic model of AI, with multiple agents from multiple sources interacting with each other, is going to force composability, and a requirement for interoperability and standardisation," he said. If he is right, whoever defines those standards early inherits the influence over the agentic economy that HTTP and TCP/IP gave to the early internet pioneers.

The insights: Cerf built the open infrastructure that made the internet's explosion possible. His final observation — that AI agents will demand the same kind of open protocols — is both a warning and a roadmap. The companies racing to build proprietary agent platforms may be building silos that the market will eventually demand to be opened.

Three ex-DeepMind researchers who built DeepStack — the first AI to defeat professional poker players at no-limit Texas Hold'em — have turned the same reinforcement learning technique toward financial markets. Their Prague-based lab, EquiLibre Technologies, is now valued at $500 million after raising an undisclosed Series A led by Creandum — the VC firm's largest single investment in one company ever. In partnership with quantitative trading firm Tower Research Capital, EquiLibre's algorithms trade billions in daily volume across the S&P 500 and Nasdaq. Since launching on crypto markets in 2025 and expanding to equities, the company claims a perfect record: zero negative months since inception.

The link between poker and trading is the same reinforcement learning logic. In both, the reward signal is unambiguous. CEO Martin Schmid puts it simply: "The nice thing about trading and markets is that the scoring is super simple — how much money did the agent make?" The company defines itself as a lab first, not a finance firm, and its advisory board includes Rich Sutton, who received the 2024 Turing Award for his foundational work on reinforcement learning. EquiLibre's founders — Schmid, Rudolf Kadlec, and Matej Moravcik — met as visiting PhD students at DeepMind's Edmonton lab, then moved back to their home country, Czechia, to build the company.

The insights: Every major hedge fund uses automation. What EquiLibre brings is a specific kind of automation trained to thrive in imperfect-information environments where incomplete data and adversarial conditions are the norm — exactly what poker and financial markets share. A $500M valuation on an undisclosed raise from a firm that called it their largest-ever bet signals that serious money believes this approach is different from what Jane Street and Renaissance already do.

Dell Technologies has announced that the majority of its server portfolio in India is now being manufactured locally, driven by accelerating AI adoption and a marked shift among Indian enterprises toward data sovereignty. The company is scaling its presence under India's Production Linked Incentive scheme and the broader Make in India initiative. The underlying driver: large Indian companies in regulated sectors — banking, healthcare, legal, public sector — are walking back from cloud-first strategies and choosing hybrid infrastructure that keeps sensitive workloads on-premises and in-country. Dell has simultaneously launched the Dell AI Factory with NVIDIA in India, its on-premise agentic AI infrastructure platform, with Dell PowerStore Elite arriving in August 2026.

This is not a supply chain move. It is a market-positioning one. As more Indian enterprises decide that data sovereignty outweighs cloud convenience, the demand for locally built, locally operated server infrastructure is growing faster than the global average. Dell's bet is that being embedded in Indian manufacturing before that inflection fully arrives is the right position.

The insights: India's cloud-first enterprise era is softening, and the shift toward sovereign, hybrid AI infrastructure is accelerating. For any company building AI products in or for India, the infrastructure story is changing — and Dell is one of the clearest signals that the enterprise incumbents have noticed.

External Affairs Minister S. Jaishankar addressed the Human Resource Mobility Forum in New Delhi yesterday, laying out India's three-pillar framework for future employment: the AI and tech economy, the green economy, and what he called the silver economy — healthcare and caregiving services for rapidly ageing global populations. "AI and automation are expected to transform labour markets... While some occupations may undergo significant change, entirely new categories of employment will also emerge," he said. On the third pillar: "Healthcare and caregiving services will become especially important with ageing populations. So, the silver economy is as important as the tech one."

India has now signed 28 Migration and Mobility Partnership Agreements with 26 countries — including Germany, Italy, and Japan — to create structured, legal pathways for Indian talent to fill global workforce gaps. The eMigrate 2.0 platform has processed over 5 million emigration clearances since its launch, and Jaishankar positioned India as an "indispensable contributor to global growth" through skilled human capital, not just manufactured goods.

The insights: For operators building in India or hiring Indian talent, the policy direction is unmistakable: India is deliberately positioning its workforce at the intersection of AI, energy transition, and elder care — the three sectors where global demand for skilled labour is growing fastest. The agreements Jaishankar is signing are the plumbing beneath that ambition.

Operator's Spotlight Read

On June 12, the Trump administration placed Anthropic's Fable 5 and Mythos 5 on the export-controlled technology list, giving the company 90 minutes' notice and forcing it to cut off all international access with no transition period. Eighteen days later, at 8 PM ET on June 30, Commerce Secretary Howard Lutnick posted on X: "Bureau of Industry and Security's evaluation of the diversion risks now presented by Claude Mythos 5 and Claude Fable 5 — the controls in the June 12 letter are withdrawn." Anthropic said it would begin restoring access on July 1 — today.

The restoration follows a staged process. Two weeks ago, Mythos 5 was restored to 100+ government-approved US organisations. Fable 5, the publicly available model, has now been fully cleared. What changed: Anthropic worked with the Commerce Department's Center for AI Standards and Innovation to implement a new safeguard that blocks the jailbreak US officials were concerned about 99% of the time. In exchange, Anthropic formally committed to three things — proactively detect and address security risks associated with the models; work with the US government on standards and protocols for Mythos, Fable, and future models; and inform the government of any malicious activity detected. Commerce Secretary Lutnick noted his office "worked closely with Anthropic to analyze and approve Fable 5 to ensure alignment across the US Government."

Notably, cybersecurity experts had argued from the start that the ban looked less like a security fix and more like leverage — a way for the administration to establish its authority over frontier AI releases. The timing of the lifting reinforces that reading: Asian competitors including Fugu and Tulonfeng were approaching Mythos-level capabilities, and the US government was under pressure to ensure American AI remained dominant in global markets. OpenAI's GPT-5.6 faced a similar government review before rollout.

The insights: Anthropic gets its models back. The US government gets something more durable: a precedent. AI labs have now accepted that the government has the authority to review, restrict, and negotiate the conditions of release for frontier models before they reach the public. The next release cycle at Anthropic — and at every lab watching this — will be built with that reality in mind. The era of launch-first, deal-with-governments-later is over.

Was this email forwarded to you? Don't miss any updates — Subscribe to TechWithAdit for sharp, no-noise tech intelligence. Stay sharp. — Adit