Last night and this morning, Nvidia released its Q2 FY2027 results. Even with a revenue base of nearly one trillion US dollars, it still delivered comprehensive results that exceeded expectations. Total revenue in Q2 reached $96.2 billion, up 106% year over year and up 18% quarter over quarter, significantly above the market forecast of $92.2 billion. As the core engine, data center revenue was $89.0 billion (92.5% of total revenue), up 117% year over year. Among this, super-large cloud providers contributed $48.7 billion. Competition such as custom ASICs could not prevent the business from accelerating. Edge computing revenue also reached $7.2 billion (up 27% year over year). Market sentiment was in intense tug-of-war: trading slightly lower before the open, falling as much as nearly 3% after the session began, then surging sharply to nearly a 5% gain. The move also lifted Micron and SanDisk after-hours, both rising nearly 4%, while SK hynix rose nearly 5%.

In Q2, adjusted gross margin was 75.0%, flat quarter over quarter. However, the company issued a third-quarter revenue outlook of $108 billion (±2%), above the $104.2 billion consensus expectation, yet it did not exceed the most optimistic estimate from some institutions at $112 billion. At the same time, the gross margin guidance was lowered to 74.0% (±0.5%), raising concerns about supply-chain costs and costs associated with ramping up the new platform, which became the main reason the stock faced pressure in early after-hours trading. Notably, in its guidance, NVIDIA does not assume any data center compute revenue from China, stripping out geographic uncertainty. NVIDIA’s non-GAAP net profit in Q2 was $53.95 billion, up 118% year over year. Free cash flow fell to $21.34 billion, down from $48.55 billion in Q1, primarily due to a $22.35 billion quarter-over-quarter increase in accounts receivable and a $5.78 billion increase in inventory tied to Rubin production ramp-up.

At the earnings call, management made an exception to provide long-term guidance early, expecting that revenue for fiscal year 2028 will grow approximately 70% year over year. They also emphasized that this figure is still constrained by supply. If it were not constrained by storage, power, contract manufacturing, and packaging limitations, the growth rate of unconstrained demand would approach 100%. ACIE business revenue spanning AI Cloud, industrial, enterprise, and sovereign AI reached $40.3 billion, up 25% quarter over quarter and up 138% year over year. The scale is now close to that of the hyperscale cloud giants. Huang Renxun noted that the unit revenue opportunity for data centers continues to evolve—from about $18 billion per GW in the Hopper era and about $25 billion per GW in the Blackwell era—rising to as much as $40 billion per GW in the Vera Rubin era. AWS also pledged to deploy an additional 2 million NVIDIA GPUs by Q2 of fiscal year 2029.

In response to market questions about the financing for its investment of nearly $50 billion in cutting-edge AI labs, management said its GPUs are general-purpose and highly liquid production assets, meaning even if a single customer leases them back, they can be quickly re-leased. NVIDIA expects its gross margin to briefly bottom out in the fourth quarter at 71%–72%, then stabilize again at 72%–73% after the higher-pricing agreement takes effect in the first quarter of fiscal year 2028. Huang Renxun summarized the current stage in one sentence: “AI has reached an inflection point; compute equals revenue.” The message conveyed by this earnings report is that the global AI capital expenditure cycle has not peaked—rather, it is in an acceleration phase of full-scale production on a new platform and large-scale deployment of enterprise-grade AI agents.