Anthropic Is Talking to the Trump…

1. Key Themes

AI Companies Are Actively Courting Government Power — Even While Under Scrutiny

Anthropic is navigating a paradox: it has been labeled a Pentagon supply-chain risk for restricting military use of Claude, yet is simultaneously pursuing a White House relationship through its new Mythos model.

"Anthropic is in talks with the Trump administration about its new Mythos model despite being labeled by the Pentagon as a supply-chain risk for placing restrictions on the military's use of its Claude models."

This signals that government contracts and regulatory positioning are becoming a primary competitive battleground for frontier AI labs, not an afterthought.

The AI Industry's Trust Deficit Is Widening — and Growing More Physically Dangerous

A Stanford HAI report and Gallup data confirm a deepening rift between AI insiders and the broader public. This disconnect is now manifesting in real-world hostility toward AI leadership.

"AI leaders have focused on managing the possibility of Artificial General Intelligence, or AGI... But everyday folks are more concerned about AI's impact on their paycheck and whether or not their power bills will go up as energy-hungry data centers are built." "The divide has been most apparent in the online reaction to the recent attacks on OpenAI CEO Sam Altman's home."

For investors and operators: public sentiment risk is no longer theoretical — it is a reputational and physical security variable.

The China AI Parity Moment Has Arrived

Stanford HAI's annual report delivers a striking geopolitical inflection point that reframes U.S. AI investment theses.

"China has effectively erased the U.S.'s lead in AI, with the two countries now trading places at the top of performance benchmarks even as the U.S. maintains an edge in capital and chips."

The U.S. edge is now narrowed to capital access and semiconductor supply — both of which are subject to policy risk — rather than model performance.

Enterprise AI Distribution Is the New Moat, and Partnerships Are Being Weaponized

OpenAI's internal memo reveals a fierce battle over enterprise distribution channels, with Amazon emerging as a breakout partner and Microsoft becoming a constraint.

"OpenAI CRO Denise Dresser told staff its new partnership with Amazon is driving 'staggering' enterprise demand while complaining that its longstanding tie-up with Microsoft has limited its ability to reach customers."

The implication: cloud distribution partnerships are now make-or-break for AI revenue scale — and locking into the wrong partner can cap growth.

Agentic AI Is Crossing Into Real Infrastructure Metrics

Vercel's IPO readiness data point is notable not just as a liquidity signal, but as evidence that AI agents are now measurably reshaping platform usage.

"ARR climbing to about $340 million and roughly 30% of apps on its platform now generated by agents."

Roughly one-third of Vercel's platform output is now agent-driven — a concrete, monetizable signal of the agentic shift moving beyond hype.

2. Contrarian Perspectives

Anthropic's Reported Revenue May Be Substantially Inflated

The consensus treats Anthropic as a strong revenue grower competing neck-and-neck with OpenAI. OpenAI's own CRO disputes this, alleging that Anthropic's figures are methodologically misleading.

"[Dresser] accused Anthropic of inflating its reported revenue by billions through accounting that 'grosses up' partner sales."

If true, this has significant implications for Anthropic's valuation multiples and for how investors should model AI lab revenue comparables across the sector. "Grossing up" partner sales is a known accounting technique that inflates topline without reflecting true economic ownership of revenue.

Gen Z Is the Most AI-Skeptical Cohort — Despite Being Among Its Heaviest Users

The prevailing narrative is that younger generations are digital natives enthusiastically adopting AI. The Gallup data inverts this.

"Young people were growing less hopeful and more angry about the technology, even though around half of the demographic was using AI either daily or weekly."

High usage does not equal high trust or brand affinity. Consumer AI companies targeting Gen Z may face a structural loyalty problem even as engagement metrics look strong.

AI Is Increasing — Not Decreasing — Legal Costs

The dominant narrative holds that AI will deflate professional services costs. The legal sector is surfacing an early counter-data point.

"Lawyers say a surge in AI-generated emails from clients is increasing workloads and could push fees higher as firms spend more time reviewing and correcting chatbot output."

Rather than compressing legal spend, AI may be generating a new category of billable review work — a cautionary signal for "AI will kill [professional service X]" investment theses.

3. Companies Identified

Anthropic AI safety company and frontier model developer Mentioned for: Simultaneous Pentagon risk designation and White House engagement over new Mythos model; accused by OpenAI of inflating revenue

"Anthropic is in talks with the Trump administration about its new Mythos model despite being labeled by the Pentagon as a supply-chain risk."

OpenAI Leading AI lab Mentioned for: Amazon partnership driving enterprise demand; Microsoft partnership described as limiting; acqui-hire of Hiro; Altman home attack incidents

"Its new partnership with Amazon is driving 'staggering' enterprise demand while complaining that its longstanding tie-up with Microsoft has limited its ability to reach customers."

Vercel 10-year-old San Francisco web platform startup Mentioned for: IPO readiness signal; $340M ARR; 30% of apps now agent-generated

"'Ready and getting more ready every day' for an IPO, CEO Guillermo Rauch said, with ARR climbing to about $340 million and roughly 30% of apps on its platform now generated by agents."

Slate Auto 4-year-old Troy, MI EV startup backed by Jeff Bezos Mentioned for: $650M Series C, $1.4B total raised; building low-cost electric pickup trucks

"Raised a $650 million Series C round led by TWG Global. The company has raised a total of $1.4 billion."

Hiro 3-year-old San Francisco AI personal finance startup Mentioned for: Acqui-hired by OpenAI; backed by Ribbit, General Catalyst, Restive

"OpenAI has acqui-hired Hiro, a three-year-old San Francisco startup that builds an AI-powered personal CFO to help users manage their finances."

pH7 Technologies 6-year-old Vancouver cleantech startup Mentioned for: $32M Series B; electrochemical metal extraction from low-grade ores and waste streams — a signal investment in critical minerals / sustainable mining

"Uses an electrochemical process to extract metals from low-grade ores and waste streams."

Ultralight 3-year-old New York healthcare AI startup Mentioned for: $9.3M pre-seed; runs clinical workflows and patient records while handling administrative tasks in the background

"Raised a $9.3 million pre-seed round led by The General Partnership."

World Liberty Financial Trump-linked crypto project Mentioned for: Investor revolt over alleged insider fund-freeze powers and preferential token unlock terms

"Facing an investor revolt led in part by crypto entrepreneur Justin Sun, who accused it of giving insiders the power to freeze funds and potentially cash out ahead of a token unlock."

Anodot Data monitoring company Mentioned for: Breached by ShinyHunters gang; stolen authentication tokens used to access downstream corporate cloud data across more than a dozen companies

"The ShinyHunters gang is demanding a ransom from more than a dozen companies after breaching Anodot and stealing authentication tokens used to access corporate cloud data."

Meta Social media and AI conglomerate Mentioned for: Building an AI avatar of Mark Zuckerberg trained on his voice, mannerisms, and public statements for internal employee interaction

"Meta is building an AI version of Mark Zuckerberg that can interact with employees using a model trained on his voice, mannerisms, and public statements."

4. People Identified

Denise Dresser — CRO, OpenAI Mentioned for: Authoring internal memo revealing Amazon partnership performance, criticizing Microsoft relationship constraints, and accusing Anthropic of revenue inflation

"OpenAI CRO Denise Dresser told staff its new partnership with Amazon is driving 'staggering' enterprise demand... [and] accused Anthropic of inflating its reported revenue by billions."

Guillermo Rauch — CEO, Vercel Mentioned for: Public IPO readiness statement with concrete ARR and agentic usage metrics

"'Ready and getting more ready every day' for an IPO."

Sam Altman — CEO, OpenAI Mentioned for: Target of multiple physical threats — a Molotov cocktail attack and a separate shooting incident near his home

"A 20-year-old Texas man was charged with throwing a homemade bomb at the San Francisco home of OpenAI CEO Sam Altman."

Justin Sun — Crypto entrepreneur Mentioned for: Leading investor revolt against World Liberty Financial despite having previously received favorable treatment from the Trump administration

"Accused it of giving insiders the power to freeze funds and potentially cash out ahead of a token unlock."

Mark Zuckerberg — CEO, Meta Mentioned for: Subject of an internal AI avatar project being built to interact with Meta employees at scale

"Meta is building an AI version of Mark Zuckerberg that can interact with employees using a model trained on his voice, mannerisms, and public statements."

Daniel Moreno-Gama — Alleged attacker Mentioned for: 20-year-old Texan charged with throwing a Molotov cocktail at Sam Altman's home; carried an "anti-AI" document listing other tech CEOs and investors

"Also carried an 'anti-AI' document listing other tech CEOs and investors, according to federal prosecutors."

5. Operating Insights

Revenue Accounting Methodology Is Now a Competitive Weapon

OpenAI's public accusation about Anthropic's revenue reporting is a warning for any startup benchmarking against AI lab comparables. Gross-up accounting — counting partner-originated revenue at full face value — can dramatically distort competitive positioning narratives used in fundraising and enterprise sales.

"Accused Anthropic of inflating its reported revenue by billions through accounting that 'grosses up' partner sales."

Takeaway for operators: Audit how your own revenue is reported relative to competitors. In fundraising, understand whether the comparable companies your investors cite are using consistent accounting methods — or you may be benchmarking against an inflated number.

Distribution Partnership Selection Has Long-Term Revenue Consequences

OpenAI's internal tension between its Amazon and Microsoft relationships illustrates a common founder trap: early distribution deals that accelerate initial growth can create structural customer access constraints later.

"Its longstanding tie-up with Microsoft has limited its ability to reach customers."

Takeaway for operators: Before signing exclusive or preferential distribution agreements with large cloud or platform partners, model the ceiling — not just the floor. Who owns the customer relationship? What segments are foreclosed?

AI-Generated Client Output Is Creating Downstream Professional Review Costs

For B2B SaaS companies selling AI writing or communication tools into professional services workflows, the productivity narrative may need revision.

"A surge in AI-generated emails from clients is increasing workloads and could push fees higher as firms spend more time reviewing and correcting chatbot output."

Takeaway for operators: If your product generates output that feeds into a regulated or high-stakes professional review process, build quality and accuracy into your core value proposition — or risk being positioned as a cost creator rather than a cost reducer.

6. Overlooked Insights

The ShinyHunters Supply-Chain Attack Vector Is a Systemic Enterprise Risk

The breach of Anodot — not a household name — enabled downstream attacks on more than a dozen large companies including Rockstar Games. This is a textbook third-party authentication token supply-chain compromise.

"The ShinyHunters gang is demanding a ransom from more than a dozen companies after breaching Anodot and stealing authentication tokens used to access corporate cloud data."

Why it matters: Most enterprise security focus is on direct perimeter defense. This incident illustrates that a mid-tier SaaS vendor with cloud authentication access can become the attack surface for the entire customer base. Investors in enterprise software and CISOs should be pressure-testing vendor authentication hygiene as a diligence criterion.

Prediction Markets Are Changing Information Consumption Behavior at Scale

A brief cultural note in the newsletter's "Detours" section may signal a larger behavioral shift with product implications.

"The 'monitoring the situation' meme... has become shorthand for compulsively tracking news and markets as prediction trading pushes more people to stay constantly plugged in."

Why it matters: Prediction markets are not just a financial product — they are creating a new class of compulsive real-time information consumer. This has upstream implications for news, data, and analytics products seeking engagement from a financially motivated audience that now has a direct monetary stake in staying informed.