Nvidia will report fiscal 2027 second-quarter results after the U.S. market closes on Wednesday, and the release is shaping up as a key test for the next phase of the AI trade rather than a routine earnings event.
Wall Street consensus calls for roughly $92 billion in quarterly revenue, nearly double from a year earlier, adjusted earnings per share of $2.09, and about $85.4 billion in data center revenue. Jefferies expects Nvidia to top that mark with $95 billion in revenue. Citi has also raised its estimates and maintained a $300 price target.
Still, the market is asking for more than a headline beat. Goldman Sachs warned that after Nvidia shares gained more than 12% in August, a combination of better-than-expected results and higher guidance may not be enough on its own to push the stock higher. In Goldman’s view, investors also want signs of improving profitability at hyperscale cloud providers, manageable customer financing exposure, and large-scale capital returns.
Options pricing points to a restrained setup
That caution is visible in derivatives markets. According to ORATS, implied post-earnings volatility for Nvidia stands at about 5.4%, the lowest level since August 2021 and well below the 7.4% average seen over the past 12 quarters.
Matt Amberson, founder of ORATS, said actual stock moves came in below implied levels in each of the past eight earnings reports. 「Options buyers have consistently overpaid for event premium, and pricing has started to correct,」 he said. For investors still constructive on the AI theme, that lowers the cost of using options to gain exposure. It also leaves room for a sharper move if management fails to address longer-term concerns that matter more than this quarter’s numbers.
Analysts broadly like the quarter, but not the immediate stock reaction
Sell-side firms are mostly upbeat on near-term fundamentals. The split is over what happens after the report.
Jefferies expects July-quarter revenue of $95 billion and October-quarter guidance of about $108 billion, both above consensus estimates of roughly $91.9 billion and $103.7 billion. Morgan Stanley is more cautious, projecting about $91.2 billion for the July quarter and $102.3 billion for the October quarter, though it still believes sequential revenue could keep rising in increments of roughly $10 billion per quarter.
Citi kept its buy rating and $300 target price, saying Nvidia has locked in HBM memory supply for 2026 through 2027, giving it support on delivery stability. Morgan Stanley maintained its overweight rating and top-pick status with a $288 target, arguing that Nvidia at roughly 17x FY28 earnings is not expensive versus its own history, semiconductor peers, or the broader market.
At the same time, Morgan Stanley noted that Nvidia shares fell the day after earnings in each of the past four quarters despite strong reported numbers. In other words, the market may already be treating an earnings beat as the base case.
Prediction market data tell a similar story. The probability of Nvidia beating expectations is priced at 96%, while the probability of the stock closing above $220 after earnings is only 43%.
Rubin is emerging as the next valuation debate
In the near term, demand for Blackwell remains strong. Better cloud trends at Microsoft and Amazon, along with long-term compute commitments from OpenAI and SpaceX, continue to support confidence in Nvidia’s data center business. But investors are already looking past Blackwell. A central question for the earnings call is whether the Rubin platform can open a new leg of growth from a much higher base.
Jefferies said Rubin could ramp materially faster than Blackwell did in its early stage. The firm’s argument rests on continuity in system design: Rubin keeps the 72-GPU Oberon NVL72 rack format, which allows Nvidia and its partners to reuse liquid-cooled data center deployment and manufacturing experience built around GB200 and GB300. The new design also sharply reduces manual cabling and piping work. Management previously said compute tray assembly time could fall from around two hours to five minutes.
That matters because it points to Nvidia’s push to turn AI hardware into a repeatable, scalable delivery model described as an 「AI factory system.」 Jefferies expects Rubin/R200 to account for about 12% of Nvidia GPU revenue in fiscal 2027 third quarter, more than 40% in the fourth quarter, and to become the primary revenue driver in fiscal 2028 first quarter. If that path holds, the market’s framing could shift from asking how long the Blackwell cycle lasts to asking whether Rubin can support a larger growth wave.
Customer financing is one of the most contested issues
The most sensitive topic in this report may be the increasingly complex financing arrangements tied to hyperscale customers.
According to Jefferies, Nvidia has reached an arrangement with SB Energy under which SB Energy will build and operate the PORTS-Pike Technology Campus in Ohio and lease it to OpenAI under a 20-year agreement, with Nvidia serving as the exclusive compute supplier. Nvidia will provide credit support for the first 4.25GW of IT capacity, hold options on the remaining 3.75GW of capacity, and invest $1.5 billion in SB Energy.
Nvidia said each generation of systems deployed at the site would correspond to around 1.5 million GPUs and generate $150 billion to $200 billion in revenue. OpenAI’s current and planned commitments total roughly 12GW and could be expanded to 16GW, implying about $600 billion in Nvidia compute purchases before 2030.
That scale has sharpened questions around what some firms describe as supplier circular financing. Morgan Stanley said the five largest hyperscale cloud providers, together with Nvidia and Broadcom, have committed more than $3.1 trillion through guarantees, leases, and other forms of financial support to drive AI infrastructure spending by other companies. Goldman Sachs and others are pressing on how much of the resulting demand is true customer capex and how much is being induced by Nvidia through credit support or direct investment.
Morgan Stanley said financing, market share, and gross margin are the three biggest points of debate for investors right now. If management does not clearly address contingent obligations, residual value commitments, and related accounting disclosures, concerns about using the balance sheet to buy growth are unlikely to fade.
Gross margin and power supply remain pressure points
Gross margin remains one of the pillars behind Nvidia’s premium valuation. Morgan Stanley expects the company to stick with its description of fiscal 2027 gross margin in the 「mid-70%」 range, suggesting near-term pressure is manageable.
Longer term, the picture is less clean. Morgan Stanley said costs are still rising across HBM, DRAM, advanced packaging, substrates, and front-end wafers. New product ramps also tend to bring yield issues and supply-chain friction. The firm added that consensus forecasts for a clear rebound in gross margin in the second half of fiscal 2028 may be somewhat optimistic.
Nvidia has also informed large customers including Microsoft, Google, and Oracle that server prices for systems using Vera Rubin and Grace Blackwell chips will rise by more than 15% early next year, with higher HBM memory costs cited as the main driver.
Power is another constraint that is getting harder to ignore. Jefferies said the gap between chip demand expectations and grid supply capacity is widening. Large cloud operators are increasingly working with utilities to build dedicated gas-fired power plants, while frontier AI labs are relying more on behind-the-meter power arrangements to bridge the time gap between grid access and compute deployment.
Gas turbines, backup generation, and fuel-cell solutions are all being pursued, but delivery timelines for equipment, interconnection approvals, and local regulation can all affect the pace of data center construction. That makes the conversion from orders to revenue less linear than the market may prefer.
Competition is adding a structural market-share question
Beyond the macro AI narrative, Nvidia’s roughly 90% share of the AI accelerator market is facing pressure from multiple directions. Google and Meta are pushing ahead with in-house AI chips. According to reports cited in the source material, 「Jalapeño,」 an inference chip developed jointly by OpenAI and Broadcom, has outperformed Nvidia Blackwell in some benchmark tests. Cerebras Systems and other newer chip companies that listed this year are also trying to win share.
Bespoke Investment Group said Nvidia’s stock performance has become notably detached from the broader semiconductor group, adding that semiconductor shares have at times moved in the opposite direction over the past several months. The PHLX Semiconductor Index has remained weak since its late-June high, while Nvidia’s prior seven-session losing streak was the longest since the current bull market began in October 2022.
That leaves this earnings report carrying more weight than a single quarter’s numbers. Investors will be listening closely not only for revenue and guidance, but also for how management frames competition, financing exposure, margin durability, and the practical limits of scaling AI infrastructure.