Anthropic's growth is exposing AI's core problem: compute costs.

Why it matters: The closer AI labs get to IPOs, the harder it becomes to hide a structural margin problem: The more customers they win, the more they spend on the compute to serve them.

State of play: Anthropic's server capacity isn't keeping pace with demand, leaving paying customers stuck on usage limits and outages.

• Server capacity and compute power are finite resources that AI labs often have to purchase before they know how much demand they'll have from customers.
• Buy too much expensive capacity, erode your margins. Buy too little and you can't meet customer demand, and they'll run to your competitors as a result.

What they're saying: Anthropic CEO Dario Amodei said there's "no hedge on earth" against overbuying compute. Buying too much would bankrupt the company if demand falls short.

• Dylan Patel of SemiAnalysis warned Anthropic may be pushed toward lower-quality compute as OpenAI locks up premium supply.
• When Anthropic capped usage during peak hours, OpenAI said it would double limits.
• Amodei has signaled he'd rather lose customers in the short term than overbuy compute and torch his margins.

Between the lines: Compute isn't just fueling customer usage. Labs also need it to train upcoming models.

• Anthropic schedules training around peak hours to reduce costs, according to a source familiar with the matter.
• "Always watch the compute, other things matter, but any new capability breakthrough probably came from throwing more compute at it," Peter Gostev, AI capability lead at Arena AI, wrote on X.

Yes, but: Compute costs are plummeting as efficiency in chips and software increases.

• But usage is skyrocketing faster so total spending keeps climbing: Classic Jevons Paradox.

Zoom out: AI capex from the hyperscalers is expected to hit nearly $700 billion this year as they race to build capacity.

• Some of that money goes toward future capacity — data center leases, power contracts — and maintaining existing systems, rather than raw compute.
• Translation: Even at record capex levels, the industry isn't buying enough compute to meet full demand.

Follow the money: While that could dissuade developers in the tech world, it has the opposite effect on Wall Street.

• Anthropic proved its spending discipline while OpenAI spent ferociously on compute, which is now resulting in less demand for shares in Altman's company, according to Bloomberg.

The bottom line: The AI race looks less like a model competition and more like a capital allocation problem — and the winners are still TBD.

Sunfish — an AI-powered fertility platform — is introducing an egg-freezing program that uses predictive models to estimate the cost of reaching a target number of eggs.

Why it matters: Fertility treatments are a significant contributor to debt for many families in the U.S., but unlike mortgages, student debt or car payments, the treatments usually give people little transparency into what the final bill will be.

Driving the news: Sunfish's new egg-freezing program, launching today, analyzes a patient's biodata to predict how many mature eggs a person can expect to bank.

• The company will cover a second cycle at no cost if the target isn't met.

Catch up quick: Sunfish's program uses several million data points from fertility cycles to predict individual outcomes from IVF — and already "guarantees the cost of treatment to achieve the desired outcome," which usually means going home with a baby.

• Sunfish says its patients currently have a 70.8% success rate achieving pregnancy, versus a 54.3% national average.

By the numbers: "The typical IVF cycle costs about $25,000, but it takes most people two or three rounds of IVF," Angela Rastegar, Sunfish CEO, tells Axios.

• That can push IVF costs above $60,000.
• Medications, donor eggs, donor sperm and genetic testing all raise the costs.
• For many U.S. households, fertility treatment now constitutes the third or fourth largest expense, alongside mortgages, auto loans and student loans.

Between the lines: Most consumer AI for parents — chatbot tutors, AI toys and digital companions for toddlers — are impractical and potentially problematic.

• Sunfish proposes the opposite: a narrow, data-heavy platform where AI crunches millions of medical outcomes to predict its clients' success.

Yes, but: Most historical fertility data skews toward white people and the affluent — the same demographics that have had access to treatment.

• Rastegar says expanding access to care is one of the most important ways to build better, more representative data over time.

What we're watching: AI-driven health tools are drawing increased scrutiny from regulators.

• Sunfish positions its product as financial planning, not medical advice, but that distinction may be tested.

Anthropic needs to explain its recent source code leaks and changes to its internal safety policies, Rep. Josh Gottheimer (D-N.J.) wrote to CEO Dario Amodei in a letter shared exclusively with Axios.

Why it matters: Gottheimer's letter highlights the growing pressure on AI companies from Washington as their tools become embedded in defense and intelligence operations, and comes amid Anthropic's ongoing battle with the Pentagon.

Driving the news: Gottheimer, a leading House Democrat on AI and cybersecurity issues, sent the letter after source material powering Claude Code leaked this week for the second time in just over a year.

• "Claude is a critical part of our national security operations. If it is replicated, we sacrifice the competitive edge we have worked so diligently to maintain in all facets of our national security," he wrote.

Zoom in: Gottheimer opened the letter by saying he supports Anthropic in its dealings with the Pentagon, but pressed Amodei on why the company has rolled back certain internal safety protocols.

Keep reading.

• Meta is putting together a team of AI researchers to supercharge its algorithms. (Business Insider)
• Elon Musk's SpaceX filed to go public, but with most details kept from public view. (Axios)
• Nearly half of college students say the rise of AI has made them think about changing their major, according to newly released polling. (Axios)

Nobel Prize-winning economist Claudia Goldin helped WNBA players renegotiate their contracts to the tune of near 400% raises, according to the Wall Street Journal.