NEW YORK — The 8.7% surge that NVIDIA Corp. shares logged on Thursday brought the world’s most valuable company to a threshold no corporation had ever crossed: a $4 trillion market capitalization. On Friday, with Federal Reserve Chair Kevin Warsh using his Jackson Hole debut to signal that rate cuts remain a distant prospect, NVIDIA NVDA investors were left to answer a harder question — whether the blowout earnings that drove that milestone can sustain a valuation built on continued triple-digit revenue growth.
Shares held near $209.81 on Friday as the broader technology sector pulled back. The Nasdaq Composite fell 0.5% on the day and the S&P 500 gave up 0.3%, as traders absorbed Warsh’s remarks from the Kansas City Federal Reserve’s annual symposium in Jackson Hole, Wyoming. The new Fed chief committed plainly to fighting inflation and declined to signal any near-term rate reduction — a message markets read as hostile to rate-sensitive growth assets. The 10-year Treasury yield held at 4.679%.
Wednesday evening’s results delivered numbers that exceeded even the most aggressive institutional forecasts. Quarterly revenue came in at $96.2 billion against a consensus estimate of $89.1 billion — a beat of more than $7 billion. Adjusted earnings per share of $2.22 cleared the $2.09 estimate. Data center revenue reached $89 billion, up 117% year-over-year, accounting for more than 92% of total quarterly sales and representing the highest single-quarter figure for any segment in NVIDIA’s history.
“Demand is accelerating,” Chief Executive Jensen Huang told analysts on the call. He described demand for Blackwell architecture GPUs — the current generation of NVIDIA’s AI accelerators — as “insane.” Huang framed compute as revenue: a formulation reflecting the structural reality that every Blackwell chip shipped translates directly into billable AI inference and training workloads for the hyperscalers, enterprises, and sovereign AI programs purchasing them. CFO Colette Kress called the quarter “outstanding,” a word that earns no argument when applied to $63.7 billion in GAAP operating income on $96.2 billion in revenue. The company returned $26 billion to shareholders during the quarter alone.
The AI cloud, industrial, and enterprise customer segment — what Huang refers to as ACIE — contributed $40.3 billion in sales, up 138% annually. That cohort, which includes the large cloud providers and a growing roster of enterprise AI deployments, drove the quarter’s outperformance and will determine whether the third-quarter guidance of $108 billion proves conservative or optimistic. At the midpoint of the guidance range, it implies roughly 90% year-over-year revenue growth from the comparable period last year.
To execute that forecast, NVIDIA needs a steady supply chain: advanced packaging from TSMC, high-bandwidth memory from Micron Technology and SK Hynix, and complex substrates from a small number of suppliers operating near capacity. Huang said constraints are easing as Blackwell production matures through the second half of the fiscal year — and that Vera Rubin, NVIDIA’s next-generation architecture, is already in production. The compressed upgrade cycle is a deliberate strategy to reinforce customer dependence by narrowing the gap between each performance plateau and the next.
On Friday morning, a second data point on AI infrastructure spending arrived from Seattle. Amazon.com Inc. announced it would deploy an additional two million high-end NVIDIA GPUs across AWS’s global cloud infrastructure during 2027 and 2028 — doubling a commitment made as recently as March and bringing Amazon’s total contracted GPU deployment to three million units. Amazon stock climbed 3.47% Friday on the AWS-NVIDIA news, the clearest single-session indicator that AI infrastructure spending is accelerating rather than leveling off.
The competitive picture, for now, remains strongly in NVIDIA’s favor. AMD’s MI350 accelerators are shipping to select hyperscale customers but hold a fraction of the installed data center GPU base. Intel’s Gaudi platform has not registered meaningful displacement of NVIDIA in any publicly reported large-scale configuration. Custom silicon programs at Google, Microsoft, and Amazon are designed to reduce long-run vendor dependence but are not replacing Blackwell in the current spending cycle. That picture will eventually change. The earnings call did not venture to say when.
Warsh’s Jackson Hole speech introduced a policy variable that NVIDIA’s guidance cannot price. Economists now assign roughly 60% probability to a 25-basis-point rate increase at the September 16 Federal Open Market Committee meeting. Higher rates raise the cost of capital for the data center buildouts driving Blackwell demand, and compress the present value of NVIDIA’s future earnings growth in equity models. The Nasdaq’s Friday retreat reflected both concerns simultaneously.
The weekly picture for the Nasdaq remained constructive — the index gained 0.9% over the five days, its second consecutive week of gains, anchored by Thursday’s surge in NVIDIA itself. The Nasdaq’s full session breakdown and weekly performance are covered separately. NVIDIA enters the long Labor Day weekend as the only company in history to have briefly carried a $4 trillion market cap. Whether that milestone becomes a floor or a ceiling depends, in part, on a Fed chair who used Friday morning to make clear that he is not finished.
Related Posts
