Morning Note
This is NVIDIA's third $2 billion equity-plus-partnership deal in a matter of weeks: CoreWeave in January, Lumentum / Coherent in early March, and now Nebius.
What NVIDIA is actually doing is building a structural demand floor for its own chips by becoming a meaningful owner of the companies that must buy those chips. The investment creates the customer, and the customer creates durable GPU revenue. That's vertical demand engineering at work!
Nebius actually emerged from the ruins of Yandex after Russia's invasion of Ukraine, spent two years restructuring, sold its entire Russian business to a local consortium for $5.2 billion in July 2024, and relaunched as an AI cloud company carrying the international assets. Most people are treating Nebius as just another hyperscaler challenger. it's not. it's a company with genuinely unusual engineering depth (former Yandex infrastructure teams are serious), a slightly complicated geopolitical history, and now a ~8.3% nvidia shareholder backing a $16-20 billion capex plan.
One more thing to watch. NVIDIA now has equity stakes across multiple large GPU buyers simultaneously. At some point someone is going to ask whether that creates conflicts, disclosure obligations, or just makes NVIDIA's demand forecasting suspiciously accurate.
3 STORIES THAT MATTER
1. Google Closes $32B Wiz Acquisition, Red Hat and Nvidia Launch AI Factory — GENERAL AI
Google's $32B Wiz buy consolidates cloud security moat. Competitors face margin pressure as security becomes table-stakes bundled offering.
2. Geopolitical Escalation Exposes Critical Supply Chain Vulnerabilities for Tech — ENERGY
Strait of Hormuz disruption halts sulfuric acid supply, crippling metal extraction for chip/datacenter components. Supply chains lack redundancy.
3. AMD CEO to Discuss HBM Supply Cooperation with Samsung Electronics Chairman — COMPUTE
AMD secures HBM supply talks with Samsung. Chip designers reduce reliance on single-source memory, stabilizing AI accelerator costs.
ON OUR RADAR
NVIDIA Invests in Thinking Machines Lab for AI Research
NVIDIA's investment in Thinking Machines Lab is a direct move to own the research layer before rivals like Anthropic and OpenAI define next-generation architectures that could reduce GPU dependency. Paired with SK Hynix's LPDDR6 breakthrough and Nscale's $2B European infrastructure raise, the compute stack is consolidating fast — NVIDIA is buying influence over the science before the science outpaces its hardware roadmap.