Analysis
Nvidia closed out its fiscal second quarter with $96.2 billion in revenue, up 106% year over year and about 5% above Wall Street's consensus estimate, sending the stock up 8.7% on Thursday to close at $227.98. Earnings of $2.22 per share more than doubled the year-ago period, and CFO Colette Kress called it "another outstanding quarter," noting growth has now accelerated for four consecutive quarters, CNBC reported.
Data Center revenue -- the number that actually moves the AI narrative -- came in at $89.0 billion, up 117% year over year, with Blackwell Ultra deployments accelerating among hyperscale customers and Data Center Networking revenue up 138%. CEO Jensen Huang framed the quarter as a turning point on the earnings call: "AI has reached its inflection point. It's doing useful work. Its tokens are productive and profitable. Now, compute is revenue." He pointed to a broadening buyer base -- multiple frontier labs, a growing open-model ecosystem and physical AI, rather than a single dominant customer, now scaling demand in parallel.
The guidance, and the number nobody is discussing enough
Nvidia guided third-quarter revenue to $108 billion, with management explicitly assuming zero China compute -- a deliberate hedge against export-control risk that still leaves room for upside if restrictions ease. Longer term, Kress projected 70% revenue growth for fiscal 2028, sharply above prior analyst models.
Alongside the results, Nvidia disclosed $279 billion in supply and capacity commitments, plus $29 billion in cloud agreements -- figures Pulse has tracked as Nvidia's balance sheet exposure has grown alongside its revenue.
The read-through
The beat rippled through the AI-infrastructure trade immediately: Broadcom gained 4.5% and Intel rose 4.4% on chip-supply optimism, while neocloud names like CoreWeave and IREN jumped on the assumption that Nvidia's guidance validates their own capacity build-outs. That is the standard playbook now -- a strong Nvidia print lifts everyone downstream, a weak one drags them all down together, which is itself a concentration risk the market has mostly stopped pricing.
The counterweight
Guiding to $108 billion next quarter is also guiding to a number Nvidia has to hit, and the company just told investors its own future is contingent on multiple frontier labs and an open-model ecosystem it does not control continuing to scale in parallel -- a diversification story that is also an admission single-customer concentration was real. The zero-China assumption in guidance is conservative on paper but underscores how much of Nvidia's addressable market sits behind a policy decision it cannot influence. And the same week Nvidia posted this print, Marvell and Rubrik -- both AI-infrastructure-adjacent -- beat estimates and got punished by investors anyway, a sign the market's patience with "the AI buildout will pay for itself eventually" is not unlimited even for the biggest name in the trade.
Nvidia's own $279 billion in supply and capacity commitments is the number worth sitting with longer than the earnings beat: it is the company betting its own balance sheet that this quarter's demand curve holds for years, not quarters.
