AI Profits Surge, $70B Capex Surge Raises Investor Concerns, Automated Skin Exams | Diet TBPN
TL;DR
Google was the clear earnings winner — Search revenue stayed very alive at $56.4 billion, up 19% YoY, while Google Cloud jumped 63% and backlog nearly doubled to more than $460 billion, with over half expected to convert within 24 months.
Wall Street is no longer rewarding capex vibes alone — the hosts say this quarter’s real test was whether each hyperscaler could justify massive AI spending now, and Google, Amazon, and Microsoft had cleaner stories than Meta.
Microsoft’s AI story is solid but still early at enterprise scale — Copilot seats rose by 5 million to 20 million, but that’s still tiny next to 450 million paid Microsoft 365 seats, so investors are watching how fast add-on adoption can compound.
Amazon’s spend looks easier to defend because AWS can sell picks and shovels to everyone — AWS grew 28% against 25% expectations, ad revenue hit $17.2 billion, and its chips business crossed a $20 billion run rate while partnering with both OpenAI and Anthropic.
Meta’s business is strong, but its AI narrative is fuzzier — revenue grew 33% to $56.3 billion, yet the company raised capex guidance by $10 billion on both ends to $125 billion-$145 billion without the obvious cloud-style revenue offsets investors want.
The back half of the show shifts from market narratives to practical AI deployment — Jensen Huang pushes back on job-loss doom, the hosts highlight automated skin exam systems as a real workflow assist for dermatology, and ARC AGI V3 still humbles frontier models with GPT-5.5 scoring just 0.43%.
The Breakdown
Rushing in from Stripe Sessions to a four-company earnings eclipse
The hosts open breathless, joking that they had to sprint back from Stripe Sessions and didn’t get nearly deep enough on the “Coasean singularity” idea from the Collison brothers. Then they pivot hard into the rare “quadkill” day where Google, Microsoft, Amazon, and Meta all report at once — a scheduling fluke they compare to a solar eclipse.
Google shuts down the “Search is dead” thesis
Google comes out as the day’s biggest winner, with the stock up around 10% after what they call a total crush. The key point: AI chatbots have not eaten Search — Search and other revenue hit $56.4 billion, up 19% YoY — while Google Cloud surged 63% and backlog swelled past $460 billion, with more than half set to be recognized in the next two years.
Microsoft’s clean beat meets OpenAI contract reality
Microsoft posted $82.9 billion in revenue, up 18%, but the conversation quickly narrows to Azure, Copilot, and the changing OpenAI relationship. Copilot seats grew by 5 million to 20 million, which sounds good until they remind you Microsoft has 450 million paid seats total; meanwhile, Azure losing OpenAI exclusivity hurts sales leverage even as Microsoft still benefits through its OpenAI equity stake.
Amazon keeps spending because AWS is still reaccelerating
Amazon remains the “capex king,” but this time the market mostly accepts the spending because AWS growth beat expectations. Quarterly revenue hit $181.5 billion, up 17%, AWS came in at 28% growth versus 25% expected, ads generated $17.2 billion, and the chips business crossed a $20 billion run rate — giving Amazon a broader set of ways to make AI infrastructure pay off.
Meta looks like two companies at once
The hosts are almost awed by Meta’s business — $56.3 billion in revenue, up 33%, with ad impressions up 19% and average ad prices up 12% — but they explain why investors still got nervous. Unlike Google, Microsoft, or Amazon, Meta can’t point to giant cloud contracts to absorb capex, so when it raises its spending range to $125 billion-$145 billion, the market has to decide whether it’s funding ad optimization, frontier AI ambition, or just higher compute prices.
The broader lesson: AI stories are fracturing
One of the sharper takeaways is that the old “biggest capex number wins” narrative is breaking apart. Google is the full-stack AI platform, Microsoft is enterprise distribution, Amazon is infrastructure plus partnerships, and Meta is a strange mix of an elite ads machine and a high-risk option on frontier AI — so investors are increasingly demanding proof tailored to each model.
Bubble talk, George Hotz, and Jensen’s anti-doom case
The hosts push back on easy bubble comparisons by noting today’s mega-cap PE ratios — Meta at 16x, Google at 17x, Amazon at 24x, Microsoft at 25x — are nowhere near dot-com-era absurdity, even if plenty of froth exists elsewhere. Then they tee up George Hotz and Jensen Huang making a similar “white pill” point: AI may kill tasks, but not necessarily jobs, and scare-stories about wiping out radiologists or software engineers can become self-defeating if society still needs more of both.
Automated skin exams, ARC AGI struggles, and Amazon’s AI podcast absurdity
The closing segment turns into a rapid-fire demo reel: they’re genuinely excited by automated skin exam technology that could make dermatology checks more precise and accessible without replacing doctors. They also note how hard ARC AGI V3 still is — GPT-5.5 scored 0.43%, Opus 4.7 scored 0.18% — before ending on a laugh at Amazon’s new AI-generated product podcasts, imagining a future where you listen to an hour-long show just to buy paper towels.