☀️ TRENDING AI NEWS
🤖 OpenAI's reasoning model disproves a geometry conjecture unsolved since 1946 - and top mathematicians are backing the claim
🏢 Anthropic projects its first ever profitable quarter with $10.9B in Q2 revenue
🏢 Meta notifies thousands of employees of layoffs to offset ballooning AI investment costs
⚡ Nvidia posts another record quarter and Jensen Huang announces a brand-new $200B market in AI agent CPUs
Something quietly shifted in the AI landscape this week - and it has nothing to do with a chatbot update or a benchmark leaderboard shuffle. OpenAI just claimed it cracked a math problem that has been open since 1946. And the researchers who publicly embarrassed them last time they made a similar claim? They're backing it up this time. That alone would be enough for a packed newsletter. But we've also got Anthropic turning its first profit, Meta cutting thousands of jobs to pay for its AI bet, and Jensen Huang announcing a new $200B market like it's just another Tuesday. Let's get into it.
🤓 AI Trivia
Anthropic was founded as a spin-off from which major AI company?
🔢 Google DeepMind
🔢 OpenAI
🔢 Meta AI
🔢 Microsoft Research
The answer is hiding near the bottom of today's newsletter... keep scrolling. 👇
🤖 OpenAI Just Cracked an 80-Year-Old Math Problem - For Real This Time
OpenAI's reasoning model has reportedly disproved a geometry conjecture that has sat unsolved since 1946 - and crucially, the mathematicians who previously called out OpenAI for fabricating a similar result are now vouching for this one.
The Verification That Changes Everything
That's the critical detail here. OpenAI has faced past embarrassment over inflated math claims that didn't hold up to scrutiny. This time, independent mathematicians have reviewed the work and confirmed the conjecture is genuinely disproved - not hallucinated, not misunderstood.
It's a meaningful signal for anyone watching reasoning models progress. Moving from 'impressive on benchmarks' to 'solving real open problems in pure mathematics' is a different category of capability entirely. Jack Clark of Anthropic said this week he expects AI to co-author a Nobel Prize-winning discovery within 12 months - results like this make that feel less like hyperbole.
🏢 Anthropic Is About to Turn Its First Profit
Anthropic has told its investors it expects to post its first profitable quarter, projecting revenue of around $10.9 billion in Q2 2026 - more than double its previous quarter. That's a staggering growth rate for any company, let alone one that's been burning billions on compute and research.
$1.25B a Month on Compute Alone
Here's a wild detail buried in the SpaceX IPO filing: Anthropic is paying xAI $1.25 billion per month for compute. That's not a typo. Fifteen billion dollars a year flowing from one AI lab to Elon Musk's operation - a partnership that surprised the industry when it was announced and now has a price tag to match. For context on how fast this space moves, we covered Anthropic overtaking OpenAI in business adoption just last week.
The profitability milestone matters. It signals that the 'AI companies spend everything on infrastructure' narrative is starting to break. Claude is clearly finding paying customers at scale - and the enterprise momentum we've been tracking on the enterprise AI tag is accelerating fast.
🏢 Meta Is Cutting Thousands of Jobs to Pay for AI
Meta notified thousands of employees of layoffs yesterday, with management explicitly framing the cuts as a way to "offset the other investments we're making" - a candid admission that AI is expensive enough to require headcount reductions elsewhere in the business.
Mandatory Transfers and New AI Teams
It's not just layoffs. Meta is also mandating that over 7,000 workers move to new teams - including two entirely new divisions focused on AI agents and cloud infrastructure. Internal messaging reportedly made clear that "transfers aren't optional." Some employees scrambled to use up benefits like headphone stipends before the cuts hit.
The job market impact of AI is getting harder to ignore. This isn't a struggling company trimming fat - Meta is profitable and growing. The cuts are a deliberate reallocation of resources from human headcount to AI infrastructure. Expect other large tech firms to use similar framing in the months ahead.
⚡ Jensen Huang Just Found a Brand-New $200B Market
Nvidia CEO Jensen Huang announced yesterday that he's identified what he calls a "brand new" $200 billion market for the company: CPUs specifically designed for AI agents. As agents move from demos to deployed infrastructure, the compute requirements look very different from traditional GPU-heavy training workloads.
Record Quarter, Slowing Growth Forecast
Nvidia also posted another record revenue quarter - though it forecasted that growth would slow in the following period, which the markets noticed. The company also revealed it holds $43 billion in startup investments - a war chest that makes Nvidia not just a chip supplier but one of the most influential investors in the AI ecosystem. That's a lot of leverage over which companies and architectures win.
The agent CPU play is worth watching closely. If AI agents become the dominant deployment pattern - as OpenAI and Google are both betting on - the underlying chip market shifts dramatically. Nvidia is clearly positioning early.
⚠️ AI Chatbots Invented Fake Scandals During the Scottish Election
A new study from thinktank Demos found that AI chatbots including ChatGPT, Grok, and Gemini made serious errors during the recent Scottish election - fabricating fake scandals, inventing candidates who didn't exist, and in some cases giving voters the wrong date for the election itself.
Electoral Commission Calls for Legal Controls
The UK's Electoral Commission responded by calling for new legal controls over AI misinformation in election contexts. This is notable - not just a researcher warning, but a formal regulatory body saying the current situation isn't acceptable.
It connects directly to the wider AI labeling and detection challenge that's been building all week. When voters can't distinguish AI-generated campaign content from real information - and chatbots are actively producing fabricated political narratives - the stakes move well beyond academic integrity or creative industry disputes. The election technology regulatory window is closing fast.
Quick aside - if you're building a project or side business and need a website up fast, 60sec.site is an AI website builder that gets you live in under a minute. Worth bookmarking. And for more daily AI coverage, dailyinference.com has the full archive.
🌎 Trivia Reveal
The answer is B - OpenAI! Anthropic was founded in 2021 by Dario Amodei, Daniela Amodei, and several colleagues who previously worked at OpenAI. The split was partly motivated by concerns about AI safety and the direction of the company - which makes today's news about Anthropic paying OpenAI's main rival $1.25B a month for compute all the more ironic.
💬 Quick Question
Meta framed its layoffs as necessary to "offset AI investments" - a line we're probably going to hear from a lot more companies this year. Do you think that framing is honest, or is it just cover for cuts that would have happened anyway? Hit reply and tell me what you think - I read every response and I'm genuinely curious where readers land on this one.
That's it for today - a lot happening on the money and math side of AI this week. See you tomorrow with more. 👋