Our Take
Nvidia is buying stakes in AI startups at a pace that dwarfs its prior behavior, but the real news is the revenue deceleration—12% growth is not a crisis, but it exposes the limits of data center demand when every hyperscaler already has Blackwell deployed.
Why it matters
Nvidia's quarter reveals the inflection point: saturation at scale. When the company owns equity stakes in every meaningful AI firm and growth still slows, it signals the market is consolidating around proven infrastructure rather than expanding into new capacity. Customers and competitors should expect a shift from scarcity pricing to deployment lock-in.
Do this week
Infrastructure teams: audit your Nvidia contract terms and multi-year commitments before Q3 negotiations begin, since Nvidia's reduced growth forecasts may force pricing or volume concessions.
The Numbers
Nvidia reported $81.6 billion in quarterly revenue (up 20% quarter-over-quarter) and $75.2 billion in data center revenue for the quarter ending April 26, per the company filing. The company authorized an $80 billion share repurchase program. Revenue forecast for the next quarter: $91 billion, representing 12% growth (per CFO Colette Kress).
The more unexpected announcement: Nvidia's private equity holdings in non-marketable securities jumped from $22 billion at quarter start to $43 billion at quarter end. The company purchased $18.5 billion in stakes during the three-month period, compared to $649 million in the prior quarter. CEO Jensen Huang disclosed pending significant capacity buildouts with Anthropic and confirmed an earlier $30 billion commitment to OpenAI (structure unspecified).
China contributed minimally to results. H200s are approved for U.S. export but, per Kress, "we have yet to generate any revenue, and we are uncertain whether any imports will be allowed into China."
The Inflection
The equity spree and the growth slowdown tell the same story: the era of capacity scarcity is ending. Every major hyperscaler, cloud provider, and model maker has already deployed or committed to Blackwell, per Kress. Nvidia owns pieces of OpenAI, Anthropic, and likely dozens of other startups now burning through that capacity. The company is no longer waiting for demand to pull supply; it is buying vertical integration and lock-in equity.
The 12% guidance is not alarming in absolute terms. But it arrives after three quarters of 20%+ growth. In markets driven by scarcity, deceleration at Nvidia's scale signals the scarcity is over. Hyperscalers are no longer competing to access chips; they are now optimizing utilization and cost per inference. That shift favors efficiency over raw throughput. Nvidia's margin story may hold, but the volume-growth narrative has peaked.
The China non-event is also telling. Nvidia has spent months signaling risk from export restrictions and alternative architectures. The quarterly results confirm that risk has not materialized—but also that no new market has opened there. The company is growing despite a closed-off region, not because of global expansion.
What to Watch
Watch Nvidia's gross margin and ASP (average selling price) over the next two quarters. If the company maintains or grows margins despite 12% revenue growth, it means per-unit pricing held firm and competitive pressure from AMD, custom silicon, or efficiency-focused alternatives did not materialize. If margins compress, the scarcity premium is eroding faster than guidance suggests.
The $43 billion in startup equity is not passive. Nvidia now has financial upside to outcomes at OpenAI, Anthropic, and others. This creates incentives to favor portfolio companies in inference workload allocation or infrastructure roadmaps—a subtle but material shift from vendor neutrality. Teams relying on Nvidia infrastructure should understand this dynamic when negotiating custom silicon or alternative arrangements.
Expect consolidation in model-maker funding. If Nvidia is the primary infrastructure financier, independent model-makers outside Nvidia's portfolio face higher effective capital costs.