Nvidia’s latest results show the AI infrastructure boom is becoming a supply-chain and power-market story as much as a chip-sales story. The company reported fiscal 2027 second-quarter revenue of $96.22 billion, up 106% from a year earlier, and said it has committed $279 billion to supply agreements, largely for memory needed to support its Vera Rubin platform.
Data-center revenue reached $89.02 billion for the quarter ended July 26, up 117% year over year and representing roughly 93% of Nvidia’s total sales, according to the company. The concentration illustrates how thoroughly Nvidia’s business now depends on the construction of AI data centers by cloud providers, enterprise customers and specialized AI computing firms.
Nvidia forecast fiscal third-quarter revenue of $108 billion, plus or minus 2%, above the roughly $104 billion Wall Street consensus cited by the company. Vera Rubin systems entered volume production shipments during the third quarter and are expected to account for about 20% of data-center revenue in the period.
The forecast indicates that Nvidia’s near-term constraint remains the availability of components and manufacturing capacity rather than a shortage of orders. Yet the same surge in spending that has lifted Nvidia’s revenue is raising costs for the hardware needed to build AI systems, particularly high-bandwidth memory, or HBM, a specialized form of memory used alongside advanced processors.
Memory costs begin to narrow Nvidia’s margin outlook
Nvidia reported a 75% gross margin on both a GAAP and adjusted basis in the second quarter, unchanged from the prior quarter and above the 72.5% level reported a year earlier. Its outlook points to a decline: the company expects a 74% gross margin in the third quarter and 71% to 72% in the fourth quarter.
The company attributed the expected step-down to rising HBM costs that are increasing faster than the price increases it can pass along to customers. Nvidia said it had raised server prices by more than 15%, but the increase has not fully offset higher component costs.
That change places memory suppliers SK hynix, Samsung and Micron in a more influential position within the AI hardware cycle. Nvidia has historically captured an unusually large share of the economics of AI computing through its processors, software and networking products. The projected margin decline suggests that some of the value generated by the spending boom is moving upstream to suppliers of scarce memory and manufacturing inputs.
Nvidia’s supply commitments rose from $119 billion at the end of the fiscal first quarter to $279 billion at the end of the second quarter. The company linked most of the increase to memory procurement for Vera Rubin, signaling that securing components has become essential to maintaining shipment growth.
Cloud providers remain the largest source of demand
Hyperscale cloud providers generated $48.7 billion in Nvidia data-center revenue, up 102% from a year earlier, Nvidia said. AI cloud providers, industrial groups and enterprise customers contributed another $40.3 billion, up 138% year over year and 25% sequentially.
Nvidia also pointed to accelerating capital expenditure by major technology companies. Alphabet, Amazon, Meta and Microsoft spent a combined $166 billion on capital expenditures during the quarter ended June 2026, Nvidia said, up 87% from a year earlier and 27% from the preceding quarter. Their capital spending has grown by 272% over the past 10 quarters since early 2024, according to the company.
The company estimates that spending by the five largest hyperscale cloud providers could rise from about $800 billion in 2026 to roughly $1.3 trillion in 2027. Such projections support Nvidia’s sales outlook, though they also imply an increasingly intense contest for data-center construction capacity, advanced chips, memory, rack space and electricity connections.
Nvidia said China contributed less than 1% of its data-center compute revenue following export controls, and its third-quarter forecast assumes no revenue from the market. That leaves the company more reliant on demand from U.S.-based cloud groups and their global data-center expansion plans.
Data-center expansion raises a power constraint for digital infrastructure
The International Energy Agency has estimated that data centers consumed about 460 terawatt-hours of electricity globally in 2022 and projected that their consumption could exceed 1,000 terawatt-hours by the end of 2026. AI-focused facilities require particularly dense clusters of processors and supporting equipment, increasing the pressure on power grids in regions where new projects are being built.
For cryptocurrency networks, the effect is likely to be most direct for proof-of-work mining operations, which compete for industrial-scale electricity and physical data-center infrastructure. Bitcoin mining economics already depend heavily on local power prices, machine efficiency and access to long-term energy arrangements. A tighter market for generation, transmission capacity and suitable sites could raise operating costs or delay expansion for miners without secured electricity supply.
The pressure will vary substantially by location. Electricity markets are regional, and some mining facilities use curtailed power, stranded energy or contractual arrangements that are not readily available to large AI campuses. Nvidia’s results do not establish that AI data centers will displace cryptocurrency miners across the market, but they reinforce the likelihood that grid access will become more valuable in areas attracting large computing projects.
Proof-of-stake networks such as Ethereum do not rely on competitive mining hardware to validate transactions, leaving them largely insulated from the direct electricity-demand competition faced by proof-of-work chains. Their operators may still encounter higher cloud-hosting or server costs, but those expenses are structurally different from running fleets of power-intensive mining machines.
Earnings beat masks the effect of investment gains
Nvidia’s adjusted earnings per share were $2.22, ahead of the $2.09 consensus estimate. GAAP earnings per share reached $2.46 and included a $7.8 billion gain on equity investments, including holdings in Intel and SpaceX. GAAP net income totaled $59.7 billion, up 126% from a year earlier.
The company generated $21.1 billion in free cash flow and returned $26 billion to shareholders through share repurchases and dividends during the quarter. Nvidia also issued $25 billion of senior unsecured notes and reported $99 billion remaining under its repurchase authorization.
Attention before Nvidia’s next expected results in late November will center on whether the company can sustain its 74% third-quarter gross-margin target while scaling Vera Rubin shipments. The answer will depend less on demand, which remains strong in Nvidia’s guidance, than on whether the industry can supply enough advanced memory, power infrastructure and physical capacity to support the computing buildout.
Explore how institutional investors react to AI and chip volatility in this ETF market insight article.
Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.
