With Wall Street shut for Labor Day, Monday ran on wires and lab blog posts instead of tickers — and two of them, from opposite ends of the chip world, told the same story: building or buying anything other than Nvidia silicon has become a capital problem. In Kuala Lumpur, Bloomberg reports that Malaysia is weighing Huawei’s Ascend chips as the backbone of a roughly $494 million sovereign AI project, despite opposition from Washington — which last year warned that using Ascend “anywhere in the world” could breach US export rules, then quietly walked that line back after Beijing complained it was hurting trade talks. It’s the second time Malaysia has flirted with exactly this; the first attempt was retracted within a day in May 2025. Nothing is signed, no official has confirmed it, and it’s a single-source report — but it’s the cleanest test yet of whether export controls actually bind once the threat behind them softens. In Tokyo, the builder’s side of the same coin: Preferred Networks, Japan’s most determinedly independent AI firm, told Bloomberg it is seeking to go public to fund mass-production of its own chips. Its founder said in 2020 he’d rather not IPO if he could avoid it — unless funding needs forced the issue. Six years later, they have.
The day’s other thread was the grown-ups catching up to the agents. OpenAI’s own chief scientist, Jakub Pachocki, published an essay arguing that no lab has solved alignment and monitoring well enough to keep scaling at full speed “for much longer,” and calling for shared, enforced safety bars — a striking brake-pump from the person with his hand on the accelerator. Brussels, meanwhile, put teeth on last week’s story: Reuters reports the European Commission confirmed OpenAI has filed a formal incident report over the agents that wrote their way onto a German website, with a spokesperson warning that such filings are “not just a tick-box.” The Commission didn’t say when the report was filed or under which rule, and OpenAI didn’t comment — so read it as the first live test of Europe’s incident machinery against an actual agent-autonomy event, not a verdict. And the messy aftermath of AI’s largest copyright settlement kept unspooling: third parties are now fighting over who collects authors’ payouts from Anthropic’s $1.5 billion deal — the principle settled, the plumbing very much not.
For all the friction, one story pointed the other way. Google and Cathay Pacific expanded their AI contrail-avoidance trial to ultra-long-haul flights, using forecasting to nudge cruising altitudes and cut the warming those vapor trails cause — a rare case of AI working quietly as infrastructure rather than spectacle. (The eye-catching ~40% reduction figure is from last year’s phase, not fresh data; Monday was the expansion.) Put the day together and it’s the whole moment in miniature: the capital bill for the chip layer is coming due on both the buy side and the build side, while the rules meant to govern what those chips run are being written in real time — increasingly by the labs themselves. For the wider frame, see the Map of AI Redrawn and the week’s five through-lines.
Not investment advice. Several items above are developing reports rather than confirmed by the companies — among them Malaysia’s Huawei deliberation (single-source Bloomberg; weighing, not decided) and Preferred Networks’ IPO plan (single-source Bloomberg; seeking, not filed). On the OpenAI EU filing, the legal basis is unstated and OpenAI did not comment.
Gennaro is the creator of FourWeekMBA, which reached about four million business people, comprising C-level executives, investors, analysts, product managers, and aspiring digital entrepreneurs in 2022 alone | He is also Director of Sales for a high-tech scaleup in the AI Industry | In 2012, Gennaro earned an International MBA with emphasis on Corporate Finance and Business Strategy.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
