Anthropic Ban Triggers AI Political Risk Reassessment; Goldman Sachs: Frontier AI Becomes National Strategic Infrastructure Under Regulatory Control
Taylor Wilson
The U.S. government blocked foreign access to Anthropic's flagship models, and Goldman partner Bobby Molavi called it a circuit breaker — frontier AI is becoming state-regulated strategic infrastructure, forcing a systemic rethink of valuations built on pure commercial logic.
What happened — how can a government shut down a private AI company's product?
The U.S. government blocked foreign users from accessing Anthropic's flagship models on security grounds; Anthropic then voluntarily shut two models for all users.
This means → whether a commercial company's product stays online is no longer entirely its own call — the government holds a de facto kill switch.
Goldman Sachs partner Bobby Molavi framed it as "a circuit breaker being activated," arguing frontier AI is evolving into state-regulated strategic infrastructure.
In plain terms = markets used to price AI firms as pure commercial companies; now they may need to price them as half-defense-contractors.
How big is the valuation shock — what do the price signals say?
Anthropic's pre-IPO proxy contract on Hyperliquid fell roughly 3.7% to about $1,627 — a modest move on its face.
The real risk is structural: Anthropic's planned IPO targets a valuation near $965 billion; OpenAI is also weighing a public listing.
This means → once pure-play AI companies go public, political tail risk moves from theory straight into the market pricing mechanism.
Anthropic already appears on a U.S. Department of Defense supply-chain risk list — because it refused to let Claude be used for mass surveillance and fully autonomous weapons.
How is Europe responding — is "strategic autonomy" a real opportunity or just a slogan?
French Prime Minister Sébastien Lecornu said the June 12 U.S. action proves France needs "strategic autonomy" in AI, backing a shift away from Palantir toward domestic alternatives.
French cloud leader OVH Groupe hit its highest share price since 2023; domestic rival Mistral AI is negotiating a funding round that could nearly double its valuation to €20 billion.
But the gap remains vast: Mistral has raised roughly $4 billion cumulatively, versus about $186 billion for OpenAI and $161 billion for Anthropic.
In plain terms = Mistral's "sovereign" tech stack still trains on Microsoft Azure and runs on Nvidia chips — it depends on America for both money and hardware.
How is China filling the vacuum — who is grabbing the market Anthropic left behind?
Within hours of Anthropic's access restrictions, Beijing-based Zhipu released its latest open-source model; its stock surged 33% in a single day.
DeepSeek imposed a permanent 75% price cut on its newest model and has become one of the most-used AI providers on OpenRouter by token volume.
This means → Beijing's playbook — low cost + state endorsement + open overseas access — is positioning China as the direct substitute for markets Anthropic vacated.
What comes next — how does this end?
Anthropic is negotiating with Washington to lift the restrictions; if the jailbreak vulnerability is as narrow as the company insists, access could resume within weeks.
But the structural question is already on the table: when model availability is an operational risk a government can toggle on and off, how much of an AI company's global footprint — and the earnings projections behind its valuation — can survive the next political intervention?
This reflects a deeper shift: the investment thesis for frontier AI is moving from "whose model is strongest" to "whose model is least exposed to political disruption."
Content is for reference only, not financial advice.