2026-05-10
Nvidia's 2026 AI equity portfolio crosses $40B — IREN warrants and Corning option cement the customer-financier model
Nvidia 2026 AI equity commitments cross $40B. New: 30M IREN warrants ($2.1B) + $3.4B managed-cloud for 5 GW at Childress TX, plus $3.2B Corning option.
CNBC reported on May 9, 2026 that Nvidia’s 2026 AI equity commitments have crossed $40 billion. The headline anchors are the existing $30B OpenAI stake plus this week’s two new positions: an IREN warrant package (~$2.1B) paired with a $3.4B managed-cloud contract, and a $3.2B option in Corning tied to three new US optical-component factories.
What ships in the IREN deal
| Component | Size | Term |
|---|---|---|
| Warrants for IREN shares | up to 30M shares at $70 strike | 5-year |
| Implied warrant value | ~$2.1B | — |
| Managed-cloud GPU contract | $3.4B | 5-year |
| Site capacity | up to 5 GW at IREN’s Childress, TX | — |
The structural pattern: Nvidia is increasingly funding its own customers’ capex so they can buy Rubin-generation hardware. The Corning component option follows the same logic on the optical-interconnect side — Nvidia funds the supply chain for the next-gen NVLink fabric.
Why this changes capacity allocation
GPU capacity is now visibly being routed through equity-tied neoclouds rather than purely through hyperscalers. CoreWeave, IREN, Crusoe, and similar players are forming a credible second tier of GPU supply that’s independent of AWS / Azure / GCP allocation queues.
For builders shopping for H200 / Rubin reserved capacity in H2 2026, the IREN Childress build creates a real second-source alternative for 5-GW-class workloads. Warrant exercise dates are a leading signal for cluster go-live — when Nvidia exercises a tranche, the corresponding capacity goes online roughly 6-9 months later.
What it doesn’t change
Nvidia’s bargaining position with hyperscalers stays roughly the same — these neocloud deals are additive supply, not substitutional. Hyperscalers continue receiving the bulk of next-gen allocation; the equity-tied second tier is the marginal supply.
Practitioner note
If you’re a startup or mid-size company shopping for serious GPU capacity (>100 H200-equivalents) in H2 2026, the equity-tied neocloud tier is now worth a real RFP. CoreWeave is the established option; IREN-Childress will start onboarding workloads in late Q3 / early Q4 once Nvidia exercises the first warrant tranche. Watch IREN’s quarterly filings — the warrant-exercise schedule is the most reliable timing signal you can get on capacity availability. Pricing will be 10-20% above hyperscaler reserved rates but 30-40% below on-demand burst pricing, which is the sweet spot for steady production workloads.
ソース
- Nvidia tops $40B in 2026 AI equity bets — CNBC ↗
- NVIDIA × IREN partnership for 5 GW Childress site ↗
- Nvidia $2.1B IREN data-center investment — Bloomberg ↗