September 10: the infrastructure chokepoint tightens

US intelligence agencies accuse Chinese AI labs of industrial-scale model theft

The NSA, FBI and CISA issued a joint advisory accusing DeepSeek, Alibaba, Moonshot AI, MiniMax, StepFun and Z.AI of systematically "distilling" American frontier models like Claude, GPT, Gemini and Grok since 2024.

Why it matters

This is the first joint US government advisory on the issue, not just a lab complaint — agencies said distillation is "the core – not merely a supplement – of their AI development strategy". It also undercuts DeepSeek's low-cost narrative: agencies allege the firm used distillation to generate training data, contradicting its claims of building models cheaply. For enterprise buyers, it raises the stakes on vetting which vendors' models rest on questionable IP practices.

What this means for you

Expect tighter scrutiny of Chinese open-weight models in procurement and compliance reviews, even though many are free and technically strong.

Finance: The claim that DeepSeek's "trivial" compute costs were partly enabled by distillation may reopen investor debate about the real cost curve of frontier AI and capex plans tied to it.

Managers: If your team uses Qwen, Kimi, or other Chinese open-weight models, be ready to answer procurement or legal questions about provenance and IP risk.

Do this: If you're evaluating or already using Chinese open-weight models in production, flag it to legal/compliance now — this advisory raises the profile of IP-provenance risk in vendor reviews.

Signal 3/5· Pay attentionSources: US Says Alibaba, DeepSeek Have 'Systematically' Siphoned AI Models - Bloomberg, US claims Chinese AI companies' core AI strategy is distilling American models - The Register

Also worth knowing

Tools & tactics

Google commits €13 billion to Finland in its largest-ever European AI buildout

Google will invest at least €13 billion in Finnish data centers, grid upgrades and clean energy over the next two years, including new sites near Kajaani, Muhos and Vaala plus an expansion of its Hamina campus.

Why it matters & what to do

Why it matters

This is one of a string of double-digit-billion commitments hyperscalers are making to a handful of power-rich, land-available locations — Fortune notes Google joins Microsoft and TikTok in betting over $30 billion on Finland alone. Compute, energy contracts and grid capacity are increasingly locked up by three or four companies, not spread across the market independent AI builders rely on.

What this means for you

The infrastructure powering AI tools is consolidating geographically and corporately, which over time can mean fewer, pricier options for anyone not renting directly from a hyperscaler.

Engineers: If your stack depends on GCP, expect steadier EU capacity and latency near the Nordics, but don't expect that abundance to translate into cheaper compute for smaller providers competing with Google for the same power contracts.

Finance: Watch Alphabet's capex guidance and Nordic utility deals — Google's move to secure up to half of the Loviisa nuclear plant's output signals hyperscalers are now underwriting national energy infrastructure to guarantee compute supply.

Do this: Nothing to do yet — just be aware that AI infrastructure is concentrating around a few well-capitalized players and energy-rich regions.

Sources: Google deepens its commitment to Finland with a €13 billion investment in AI infrastructure, Meet the data center capital of Europe as Google joins Microsoft and TikTok in betting over $30.2 billion on Finland

Signal 2/5· Worth a glance

Tools & tactics

Anthropic locks in up to a million Google TPUs, even as it builds its own chip strategy

Anthropic is dramatically expanding its use of Google Cloud, including up to one million TPUs, in a deal worth tens of billions of dollars that will bring over a gigawatt of capacity online in 2026.

Why it matters & what to do

Why it matters

Anthropic already runs a multi-chip strategy across Google TPUs, Amazon Trainium and Nvidia GPUs, plus Amazon's giant "Project Rainier" cluster — yet it's still going deeper on Google, not less. Even the best-funded AI labs can't build their way out of needing hyperscaler infrastructure at this scale.

What this means for you

Frontier AI's bottleneck is still physical compute, not model ideas — and that compute runs through a handful of cloud giants no matter how much labs diversify.

Finance: Watch Alphabet and Amazon earnings calls closely — AI lab spending like this is becoming a material, recurring revenue line for both, not a one-off.

Managers: If your roadmap assumes ever-cheaper, ever-available AI compute, budget for continued scarcity — even Anthropic is locking in supply years ahead.

Do this: Nothing to do yet — just be aware that compute supply, not model quality, is the constraint shaping AI pricing and availability into 2026.

Source: Expanding our use of Google Cloud TPUs and Services

Signal 3/5· Pay attention

Models & capabilities

MediaTek's August sales jump 44% on AI chip demand

MediaTek reported a 44% surge in August sales to NT$64.2 billion ($2 billion), far outpacing analyst forecasts of 10% growth for the quarter, as its AI chip business with customers like Google gains momentum.

Why it matters & what to do

Why it matters

MediaTek took a $3.5 billion investment from Nvidia last week, and this is the first hard sales evidence that its pivot from smartphone chips to AI data-center silicon is paying off. It's a signal that the value from AI inference isn't just flowing to Nvidia and hyperscalers — it's also lifting the chip designers and foundry partners doing the custom silicon work underneath them.

What this means for you

When a GPU giant like Nvidia puts billions into a supplier, and that supplier's sales then jump far past expectations, it's a sign the AI buildout is spreading real revenue beyond the obvious winners.

Finance: MediaTek's results are an early read on how much of Big Tech's AI capex is actually converting into supplier revenue, not just spending announcements.

Do this: Nothing to do yet — just note MediaTek as a bellwether for how much AI infrastructure spending is trickling down to chip suppliers.

Source: Nvidia-Backed MediaTek's Sales Soar 44% With AI Chip Momentum

Signal 2/5· Worth a glance

Money & markets

Legal AI startup Harvey raises $550M, valuation nearly doubles to $15.5B in nine months

Harvey, the legal AI startup, has raised another $550 million in funding, this time at a $15.5 billion valuation, co-led by Diffusion and Lightspeed Venture Partners.

Why it matters & what to do

Why it matters

With this latest infusion of capital, the company has raised more than $1.55 billion total and nearly doubled its valuation in about nine months — a pace that makes the AI mega-round headlines look almost normal. It shows investors are willing to write nine-figure checks for a narrow, high-value vertical, not just for foundation-model giants.

What this means for you

Investors are betting that AI tools built for one profession, deeply, can be worth as much as broad platforms — a signal for anyone weighing "horizontal vs. vertical" AI bets in their own career or portfolio.

Finance: A private company reaching this valuation on the back of real revenue growth (Harvey said it hit $190 million in annualized revenue) suggests vertical AI could be the next place mega-cap-style returns show up outside public megacaps.

Managers: If a legal-AI vendor can raise at this scale, expect similar-shaped vertical tools to arrive fast in your own function — worth scouting now rather than waiting.

Do this: Nothing to do yet — just note which vertical AI tools are gaining real revenue traction in your industry, not just funding headlines.

Source: Harvey hits $15.5B valuation, months after reaching $11B — TechCrunch

Signal 3/5· Pay attention

One line to sound smart

“Frontier AI's real constraint is compute access and energy contracts, not model ideas — and three companies are locking it all up.”

Tool worth a look

Claude is the AI assistant this brief is built with — genuinely useful for drafting, summarizing dense material, and thinking through what a development actually means for you. An honest pick, not a paid link.

Try Claude →

Futureproof Daily is researched and written by AI against our editorial standards — see how we work. Sources are linked on each item. Nothing here is financial, investment, or legal advice.