Today in AI — 1 May 2026

Earnings week painted a single picture repeated four times: every hyperscaler is reporting AI-driven acceleration, and collectively they're on track to spend $700 billion on AI infrastructure this year. The revenue is real. Whether it justifies the spend is a different question — one that Zuckerberg would prefer you not ask.

The capex arms race

Google Cloud grew 63%, Azure grew 40%, AWS posted its fastest quarter in nearly four years. Microsoft's annualised AI revenue hit $37 billion, up 123%, and the company ended its revenue share payments to OpenAI. But the spending side is growing even faster. Combined hyperscaler capex will surpass $700 billion in 2026, and Meta alone raised its forecast to $125-145 billion. When asked what shareholders get for that, Zuckerberg called return on investment "a very technical question." Meta stock dropped 10%. The railroad analogy is worn out, but the pattern fits: massive infrastructure build-outs where the winners aren't always the ones laying track.

• Big Tech is about to spend $700 billion on AI this year — no one knows where it ends — Fortune
• Google Cloud crosses $20 billion in quarterly revenue as AI demand drives 63% growth — Fortune
• Microsoft cloud and AI strength fuels Q3 results with $37 billion AI run rate — Microsoft
• Meta spending up to $145 billion on AI this year — Zuckerberg calls ROI question 'very technical' — Fortune
• Amazon Q1 beats estimates as AWS posts fastest growth in 15 quarters — CNBC

The valuation ladder

The capital isn't only flowing through hyperscalers. Anthropic is closing a round that would value it above $900 billion, surpassing OpenAI. SoftBank is spinning out a new AI and robotics company called Roze targeting a $100 billion US listing. Cerebras is aiming for mid-May to become the first pure-play AI chip company on a public exchange. Private AI valuations are now competing with the market caps of established tech giants.

• Anthropic's $900 billion valuation round could close within two weeks — TechCrunch
• SoftBank plans $100 billion US IPO for new Roze AI and robotics spinout — CNBC
• Cerebras targets mid-May IPO as first pure-play AI chip company to go public — Seeking Alpha

The multi-cloud scramble

OpenAI launched on Amazon Bedrock one day after restructuring its Microsoft partnership, ending cloud exclusivity. Amazon invested $50 billion; OpenAI committed to two gigawatts of AWS Trainium chips. Meanwhile, the US-China AI split deepened on two fronts: China blocked Meta's $2 billion acquisition of Manus, and Huawei expects $12 billion in AI chip revenue as DeepSeek V4 pulls Chinese cloud orders away from Nvidia. The model layer is decoupling from infrastructure lock-in, but the hardware layer is splitting along national lines.

• OpenAI models, Codex, and Managed Agents arrive on AWS via Amazon Bedrock — OpenAI
• China blocks Meta's $2 billion acquisition of AI startup Manus — CNBC
• Huawei expects AI chip revenue to hit $12 billion in 2026 as DeepSeek V4 drives orders away from Nvidia — The Deep Dive

Supply chains and checkout lanes

OpenAI revoked its macOS app certificate after a compromised Axios library entered its signing process, a reminder that AI products inherit all the fragility of their software supply chains. On the product side, all three major labs are now racing to own agentic commerce, each building AI-powered checkout flows that could reshape how consumers buy online.

• OpenAI revokes macOS app certificate after Axios supply chain compromise — The Hacker News
• The agentic commerce race heats up as OpenAI, Google, and Anthropic compete to own the checkout — Digital Commerce 360

For builders, the signal worth tracking: hyperscaler AI revenue is accelerating, but infrastructure spend is growing faster still. That gap closes through new product categories, pricing power, or write-downs, and which one it is determines whether this capex cycle looks like cloud computing in 2015 or fibre optics in 2001.

---