The U.S. effectively anointed OpenAI as its primary state AI supplier, showering it with unprecedented capital, while pushing Anthropic into a formal penalty box for refusing the same role. Users and global enterprises are reacting in the opposite direction, driving Claude to the top of the App Store and toward a huge revenue run rate despite the ban.
Underneath it all, GPUs, cloud regions, and even data centers are turning into political assets rather than neutral infrastructure.
Key Events
/OpenAI raised $110B from Amazon, Nvidia, SoftBank and others, implying a valuation around $840B.
/Amazon committed up to $100B of AWS services over eight years to run OpenAI workloads.
/The Pentagon labeled Anthropic a 'supply‑chain risk' and Trump ordered all federal agencies to stop using its AI.
/ChatGPT uninstalls surged 295% after OpenAI’s Pentagon deal as the 'Cancel ChatGPT' movement spread.
/Anthropic’s Claude app hit #1 on the App Store with over 500k downloads in a single day.
Report
Washington just turned frontier AI into a two‑bloc system: OpenAI as the state‑aligned supplier and Anthropic as the sanctioned dissenter.
At the same time, OpenAI locked up unprecedented capital from Amazon and Nvidia while Anthropic’s 'ethics premium' is converting into real user growth and revenue.
the defense‑aligned vs ethics‑first AI split
OpenAI agreed to deploy its models inside the U.S. Department of War’s classified networks, under a contract critics say allows safeguards to be bypassed in practice.
Sam Altman publicly called the deal "opportunistic and sloppy" and admitted OpenAI has no real say over how the Pentagon uses its systems.
Anthropic walked away from a similar contract over concerns about mass surveillance and autonomous weapons, even after building a custom Claude model reportedly 1–2 generations ahead of its consumer offering for military use.
The Pentagon retaliated by designating Anthropic a supply‑chain risk, and Trump ordered all federal agencies to stop using its technology, putting it in the same penalty box as Huawei‑style vendors.
More than 300 employees across Google and OpenAI signed an open letter backing Anthropic’s refusal, while a visible "Cancel ChatGPT" movement framed OpenAI as the company that took the deal.
openai’s capital stack and resilience to backlash
Against that backdrop, OpenAI raised $110B in a single round from Amazon, Nvidia, SoftBank and others, the largest tech funding event on record.
Investors are marking the company at roughly $730B pre‑money on the back of that raise. Amazon tied the deal to a commitment to provide up to $100B of AWS services over eight years to run OpenAI workloads, effectively pre‑selling massive cloud capacity.
OpenAI reports an annualized revenue run rate near $25B, a scale that makes it look more like a quasi‑utility than a startup. Yet ChatGPT uninstalls still jumped 295% after the Pentagon deal, as users canceled subscriptions and rallied under the "Cancel ChatGPT" banner without denting investor enthusiasm.
anthropic turns sanction into demand
Anthropic, meanwhile, is approaching a $20B revenue run rate despite being banned from U.S. government systems. That run rate reportedly increased by about $5B in just a few weeks, signaling accelerating commercial pull.
Roughly 80% of Anthropic’s enterprise demand is non‑American, which blunts the impact of losing federal business and redirects growth to international and commercial buyers.
Claude’s mobile app climbed from #129 to #2 on the App Store in a month and then hit #1, including a day with over 500,000 downloads as users left ChatGPT.
Commenters increasingly describe Anthropic as the "morally superior" alternative and praise its refusal to participate in mass surveillance or autonomous weapons programs.
compute and cloud as contested chokepoints
Nvidia is recalibrating its exposure to frontier labs, with Jensen Huang saying a $30B investment in OpenAI may be its last and that Nvidia is pulling back from both OpenAI and Anthropic.
Google signed a multi‑billion‑dollar TPU deal with Meta that could capture up to 10% of Nvidia’s annual datacenter revenue, directly challenging its grip on AI compute.
Nvidia is hedging by investing $2B into Lumentum to advance optics for AI infrastructure while touting large efficiency gains from Hopper to Blackwell and Rubin GPU generations.
The U.S. government is simultaneously weighing tighter rules on AI chip exports that would further restrict advanced parts from Nvidia and AMD, making access to high‑end GPUs a regulatory issue, not just a budget line.
And Iranian drones have already damaged three AWS data centers in the UAE and exposed weaknesses in Amazon’s disaster‑recovery posture, turning hyperscale regions into visible wartime targets.
incumbent infra giants overextending into AI
Oracle is cutting thousands of employees and freezing hiring in its cloud division while planning to invest tens of billions of dollars into AI data centers and GPUs.
Executives expect Oracle’s cash flow to remain negative until at least 2030 as those bets ramp, effectively front‑loading AI capex against a very long payback window.
Palantir is leaning hard into controversial military AI work—from tracking aid deliveries in Gaza to using Claude to identify over 1,000 targets in the US‑Israeli conflict with Iran—while Peter Thiel has been selling down his stake.
CEO Alex Karp has said that people who think AI will not be nationalized are "stupid", making explicit that Palantir’s strategy assumes ever‑tighter fusion between state power and AI platforms.
What This Means
The AI stack is hardening into two poles—a state‑aligned, capital‑saturated utility centered on OpenAI and a sanctioned but fast‑growing, ethics‑positioned rival in Anthropic—while access to compute and cloud is becoming a politicized chokepoint. Every meaningful exposure to AI, cloud, or defense tech now implicitly embeds a choice about which side of that emerging order it leans toward.
On Watch
/The FCC’s approval of Charter’s $34.5B acquisition of Cox, making it the largest U.S. ISP, could reset bargaining power for last‑mile access and streaming distribution.
/Xiaomi’s humanoid robots achieving a 90.2% task success rate in factories and its Vision GT EV concept signal Chinese OEMs pushing into full‑stack robotics and electric vehicles.
/Possible new U.S. rules on AI‑chip exports by Nvidia and AMD would further entangle high‑end compute access with geopolitics and sanctions policy.
Interesting
/Trump's directive for federal agencies to stop using Anthropic's technology includes a six-month wind-down period.
/The Lawfare analysis suggests that the government's legal stance on Anthropic's designation could face severe challenges, potentially undermining its authority.
/Anthropic's Claude AI was utilized by the US military during operations in Iran, despite the subsequent ban by Trump.
/ChatGPT has reached 900 million weekly active users and 50 million paying subscribers.
/Anthropic's Claude AI is responsible for writing over 80% of the code deployed by its team, showcasing its integral role in development.
We processed 10,000+ comments and posts to generate this report.
AI-generated content. Verify critical information independently.
/OpenAI raised $110B from Amazon, Nvidia, SoftBank and others, implying a valuation around $840B.
/Amazon committed up to $100B of AWS services over eight years to run OpenAI workloads.
/The Pentagon labeled Anthropic a 'supply‑chain risk' and Trump ordered all federal agencies to stop using its AI.
/ChatGPT uninstalls surged 295% after OpenAI’s Pentagon deal as the 'Cancel ChatGPT' movement spread.
/Anthropic’s Claude app hit #1 on the App Store with over 500k downloads in a single day.
On Watch
/The FCC’s approval of Charter’s $34.5B acquisition of Cox, making it the largest U.S. ISP, could reset bargaining power for last‑mile access and streaming distribution.
/Xiaomi’s humanoid robots achieving a 90.2% task success rate in factories and its Vision GT EV concept signal Chinese OEMs pushing into full‑stack robotics and electric vehicles.
/Possible new U.S. rules on AI‑chip exports by Nvidia and AMD would further entangle high‑end compute access with geopolitics and sanctions policy.
Interesting
/Trump's directive for federal agencies to stop using Anthropic's technology includes a six-month wind-down period.
/The Lawfare analysis suggests that the government's legal stance on Anthropic's designation could face severe challenges, potentially undermining its authority.
/Anthropic's Claude AI was utilized by the US military during operations in Iran, despite the subsequent ban by Trump.
/ChatGPT has reached 900 million weekly active users and 50 million paying subscribers.
/Anthropic's Claude AI is responsible for writing over 80% of the code deployed by its team, showcasing its integral role in development.