Key points
- Nvidia (NVDA) fell for six straight sessions through Aug. 21, closing at $214.76.
- Nvidia has beaten EPS estimates for 14 straight quarters, and still fell the next day in six of its last eight reports.
- Nvidia estimates China export restrictions have eliminated about $4.6 billion a quarter in H20 sales since May.
Nvidia (NVDA) closed down for six straight sessions, from $225.30 on Aug. 13 to $214.76 on Aug. 21, a drop of about 4.7%. The stock reports fiscal second-quarter results after the close on Wednesday, Aug. 26, with the earnings call set for 5 p.m. ET.
Six red sessions before earnings
The decline marks Nvidia's longest losing streak since shares fell for seven sessions in August and September 2022. Analysts expect about $92 billion in revenue for the quarter ended July 26, up 97% from a year earlier. The consensus EPS estimate stands near $2.07.
14 straight beats, four straight drops
Nvidia has beaten EPS estimates in each of the last four quarters, by 5.42%, 6.58%, 4.84% and 3.96%. The stock fell the next trading day after every one of those reports anyway, down 0.8% in August 2025, 3.2% in November 2025, 5.5% in February 2026 and 1.8% in May 2026.
Zoom out further and the streak gets longer. Nvidia's last EPS miss came on Nov. 16, 2022, when non-GAAP earnings came in light during the same gaming and cryptocurrency inventory bust that produced the seven-day losing streak referenced above. Every quarter since, 14 straight through the report in May 2026, has beaten estimates. Limit the view to just the last eight of those reports, the ones with next-day price data available, and the stock still dropped the next session six times.
That gap between the headline beat and the stock reaction is why this report is being framed as a test of guidance, not of the quarter Nvidia already reported. "Nvidia expects upside to $500 billion in AI demand from Blackwell and Rubin," Citi analysts wrote in a note covered by Seeking Alpha ahead of the print, pointing to the next generation of Nvidia's chip platforms as the source of any surprise, up or down.
Hyperscalers plan up to $800 billion in spending
Nvidia supplies about 90% of the AI GPU market, making its results a measure of demand across the data center industry.
The largest cloud companies have raised their 2026 capital spending plans well past last year's forecasts:
- Amazon expects to spend about $200 billion, up from $125 billion.
- Google expects $175 billion to $185 billion, up from $91 billion.
- Meta expects $130 billion to $145 billion, up from $72 billion.
- Microsoft expects $110 billion to $120 billion, up from $90 billion.
The five largest hyperscalers expect to spend a combined $775 billion to $800 billion on AI infrastructure in 2026.
Nvidia founder and CEO Jensen Huang said the estimated $700 billion already committed by big technology companies represents "just the start of something far bigger," according to comments Fortune reported in February. The Aug. 26 report gives investors a full quarter of sales to check against those spending plans.
The China hole in the numbers
Nvidia has booked zero China data center compute revenue since its May 2026 results, after export restrictions cut off the H20 chip built specifically for that market. The company has estimated the restrictions cost about $4.6 billion a quarter in lost H20 sales, a gap it's offset so far with demand from everywhere else. Washington approved H20 exports to China again in recent months, but Nvidia hasn't reported actual orders behind that approval yet. Whether China shows up as a revenue line, or stays at zero with just a mention on the call, is one of the more binary outcomes in this report.
Why the whole AI trade is watching one earnings call
Because Nvidia sits at the center of the AI supply chain, its guidance tends to move stocks well outside its own ticker. A strong data center number and confident capex commentary would likely lift chip suppliers like AMD, Broadcom (AVGO) and Marvell (MRVL), memory makers like Micron (MU) that sell into the same server buildout, and the hyperscalers themselves, whose stock prices already reflect the spending totals above. A soft guide, or hedged commentary on China or margins, would cut the other way across the same group, since the entire AI trade has been pricing in Nvidia's growth rate as a proxy for the buildout's health.
Nvidia has not missed an earnings estimate since 2022. The question this time is whether a beat is still enough.
