Nvidia Corp. has unleashed a torrent of cash on its AI empire. In mere months, the chip powerhouse committed over $36 billion to partners and suppliers. That’s more than Iceland’s yearly economic output. The moves lock in capacity. They guarantee chips, optics, and compute for the next wave of artificial intelligence.
Start with OpenAI. Nvidia finalized a $30 billion equity stake in the ChatGPT creator. This replaced a prior non-binding $100 billion data-center pledge. The investment came amid OpenAI’s $110 billion raise, valuing it at $730 billion pre-money, as Reuters reported. CEO Jensen Huang called it potentially “the last time” before an IPO. “Might be the last,” he said in a CNBC interview, signaling Nvidia’s pivot from pure supplier to shareholder (Yahoo Finance).
But OpenAI was just the opener. Nvidia followed with $2 billion checks across the board. First, photonics firms Lumentum Holdings Inc. and Coherent Corp. Each got $2 billion, plus multibillion-dollar purchase commitments for lasers and optical tech vital to data-center networking. Nvidia’s press release detailed U.S. manufacturing expansions and R&D boosts (Nvidia Newsroom). Lumentum’s CEO later projected AI orders filling books through 2028 (Bloomberg).
Marvell Technology Inc. pocketed $2 billion next. The deal fused Marvell’s custom AI chips and networking into Nvidia’s NVLink Fusion platform. Silicon photonics collaboration promises faster data flows in AI factories, per Reuters via Yahoo Finance. Marvell shares jumped 13% on the news (CNBC).
Nebius Group NV, an AI cloud provider, scored another $2 billion. It plans Europe’s largest data centers, fueling hyperscale growth (CNBC). And Thinking Machines Lab, founded by ex-OpenAI CTO Mira Murati, landed a gigawatt of Nvidia’s Vera Rubin chips plus “significant” funding (Yahoo Finance).
Chip Shortages Force Preemptive Strikes
Why the frenzy? Supply can’t keep pace. Hyperscalers plan $700 billion in AI capex this year. Memory alone claims 30%, up fourfold since 2023, per SemiAnalysis. Nvidia snags “very very preferred” DRAM pricing, leaving others scrambling.
TSMC, Nvidia’s foundry, faces the crush. Nvidia reportedly shelled out $10 billion in prepayments—call them advances—to secure capacity amid shortages, as CNBC’s Kristina Partsinevelos noted. TSMC’s 2026 capex hits $54 billion midpoint, doubling down on CoWoS packaging Nvidia dominates. Dylan Patel of SemiAnalysis warns Nvidia locked up allocations early; rivals like Google squeeze in late.
ASML Holding NV, the EUV lithography monopoly, enables it all. Its machines, costing $300-400 million each, underpin TSMC’s advanced nodes for Nvidia GPUs. ASML raised 2026 guidance to $43 billion on surging AI demand.
Huang dismisses slowdown fears. Nvidia’s latest quarter smashed estimates, with data-center sales soaring. Big Tech’s spending—Amazon, Google, Meta, Microsoft—nears $400 billion yearly, mostly Nvidia-bound. Yet rivals stir. Meta deploys in-house MTIA chips post-millions of Nvidia GPUs (CNBC). OpenAI taps Broadcom for custom silicon.
Still, Nvidia rules. It captures 80-90% of AI accelerators. Investments de-risk the chain. They expand U.S. optics fabs. They cement ecosystem lock-in via NVLink.
Risks loom. Debt fuels some capex. Regulators eye Nvidia’s dominance—U.S. probes loom. China export curbs cost billions. But demand endures. BlackRock’s $30 billion semiconductor ETF (SOXX) climbed 40% in April alone, led by Broadcom, Micron, AMD, Nvidia.
And so the blitz continues. Nvidia isn’t just selling chips. It’s buying the future. One $2 billion bet at a time.
WebProNews is an iEntry Publication