Tech • AI • Robotics • Game
THE AFTER: OpenAI Challenges Math, BlaBlaCar Turns to AI & French Tech Takes a Hit!
9.3/10TL;DR
Big Tech is accelerating an unprecedented AI infrastructure spending cycle while a French entrepreneur is testing how AI agents could rewrite and stress-test parts of the French legal code.
KEY POINTS
A record AI spending boom
Microsoft, Google, Amazon and Meta are pouring roughly $700 billion into data centers and related AI infrastructure this year, with expectations that spending could rise further next year. Some observers now compare the scale of investment to the largest infrastructure waves in modern economic history, from railways to electrification, with estimates that more than 3% of US GDP could be absorbed annually by AI build-out through 2032.
Google faces investor anxiety despite cloud growth
Alphabet came under pressure after results revived concerns about whether its cash generation can keep pace with AI capital spending. The company’s stock fell sharply after earnings, while discussion centered on a slowdown in its core search and advertising growth and a negative free cash flow figure driven by infrastructure expansion. At the same time, its cloud business remains a major bright spot, with strong growth reinforcing the idea that Google is trying to reinvent itself rather than rely only on legacy ad revenues.
Debt and data centers become central to the AI race
The AI boom is increasingly financed not only by operating cash but also by debt, raising questions about how long the current pace can continue. Even so, demand for computing capacity remains intense, and major technology companies appear determined to keep building. The strategic logic is clear: whoever controls the compute, chips, cloud capacity and model stack could dominate the next phase of the digital economy.
Google’s long game runs through DeepMind and custom chips
Beyond short-term market reactions, Google DeepMind remains one of Alphabet’s most important strategic assets. Its work in fundamental research, health applications and model development is paired with in-house TPUs, Google’s AI chips designed to reduce dependence on Nvidia and improve efficiency inside its data centers. Supporters argue that this combination of research, infrastructure and products positions Google strongly for an “internet of agents” in which machine-to-machine search, shopping and transactions could become a vast new market.
OpenAI claims scientific progress in three fields
OpenAI said it had produced new results in mathematics, computer science and physics. The most discussed claim involved complexity theory, where the company presented work touching the boundary between problems considered intractable and those that might be solved more efficiently than previously assumed. The immediate practical consequences remain unclear, but the announcement fueled debate over whether AI systems are beginning to contribute meaningfully to frontier scientific research rather than only summarizing known knowledge.
AI enters the French legal system by experiment
A young French entrepreneur based in San Francisco, Alex Massme, has used AI agents to process about 30,000 legal entries focused mainly on economic texts, including the labor code and urban planning code. His goal is not wholesale ideological deregulation but the detection of contradictions, outdated references and drafting errors that accumulate in a large legal system. He argues that work that might cost legal teams millions of euros can now be prototyped for a few thousand euros in computing credits.
From legal drafting to “red teaming” the law
The project is now shifting from a one-night demonstration to a more structured effort with collaborators preparing a scientific publication. The approach borrows from cybersecurity: “red team” the law by looking for exploitable loopholes, inconsistencies and edge cases from the perspective of employers, employees, entrepreneurs or litigants. Supporters say such methods could make the legal code more understandable and auditable, while critics would likely note the need for extensive human verification before any policy use.
A broader shift from public ecosystems to private initiative
Alongside the debate over AI, frustration surfaced around cuts to French Tech support and startup community funding. The underlying argument is that if public money for networking and ecosystem-building shrinks, founders and private companies may increasingly organize events, communities and support structures themselves. That same do-it-yourself logic now appears in legal experimentation, where private actors are using AI to test what public institutions have not yet attempted at scale.
CONCLUSION
The week highlighted two fronts of the AI transition: a colossal industrial race for compute in the United States and early experiments in using AI to refactor public systems such as law in France. Together, they suggest that AI is moving from hype to infrastructure and from chat interfaces to institutional redesign.
Ask a question