Nvidia reports August 26 with the Street expecting $91B in Q2 revenue. Here's what the data, prediction markets, and AMD threat mean for traders.
August 20, 2026
Key Points
Nvidia's Q1 FY2027 non-GAAP EPS hit $1.87 on $81.61B revenue (+85.23% YoY), and management guided Q2 to $91B — a 95% year-over-year increase if met.
Prediction markets are pricing only a 49.5% probability that NVDA closes August above $220, signaling the Street is hedging into the print despite bullish fundamentals.
Traders should watch whether Nvidia confirms or walks back its $250B OpenAI data center guarantee on the August 26 call — that single line item could move the stock more than the headline EPS.
Nvidia closes Thursday at $219.74, sitting roughly 28% below its 52-week high of $236.26, with six days until the most consequential AI earnings print of the year. The number that matters: $91 billion in Q2 revenue — a figure management itself set as guidance after Q1 came in at $81.61 billion on 85.23% year-over-year growth. Meeting it would represent a 95% annual revenue increase. Missing it, even narrowly, into a market already pricing uncertainty at a coin-flip probability, could be ugly.
What the Q1 Print Actually Promised
The Q1 FY2027 report filed May 20 was not just a beat — it was a structural statement about where AI infrastructure spending is going. Non-GAAP EPS of $1.87 cleared the $1.7738 consensus. Data Center revenue alone reached $75.25 billion, with networking — the switches, cables, and interconnects that tie GPU clusters together — up 199% year-over-year. That networking number matters because it shows hyperscalers are not just buying GPUs in isolation; they are building out full-stack AI infrastructure at a pace that generates attach-rate revenue across Nvidia's entire product portfolio.
Management did not stop at reporting results. They raised supply commitments to $145 billion, a figure that functions less as a forecast and more as a signal to customers: Nvidia is locking in production capacity at TSMC, and anyone not in that queue risks falling behind on AI buildout timelines. TSMC's own Q2 numbers — 35.64% revenue growth year-over-year, gross margins at 67.7%, and $29.44 billion in approved capex for further expansion — confirm that the foundry side of the trade is keeping pace. The supply chain is tight, but it is not broken. TSMC's financials represent the clearest independent corroboration that AI chip demand is not softening heading into Nvidia's August 26 print.
The guidance mechanism itself deserves scrutiny. When Nvidia guided $91 billion for Q2, they were embedding approximately 12% sequential growth over an already record quarter. Sequential growth of that magnitude, sustained across multiple quarters, is historically rare at this revenue scale. The bears' case is not that AI demand is collapsing — it is that $91 billion requires a near-perfect execution environment across supply, customer delivery, and hyperscaler absorption of new GPU capacity simultaneously.
The AMD Variable and What It Costs to Ignore It
Jefferies' readout on AMD's Advancing AI 2026 event introduced a complication bulls did not want six days before Nvidia's print. AMD's Helios platform, paired with the ROCm.AI software stack, is gaining traction precisely where Nvidia is most exposed: enterprise and hyperscaler workloads where customers have strong incentives to avoid single-vendor dependency. Anthropic has committed to a 2GW Helios deal. OpenAI has a 6GW AMD commitment on the books. Those are not pilot programs — they are infrastructure-scale procurement decisions.
AMD's Q2 2026 data center revenue hit $6.72 billion, up 107% year-over-year. That growth rate numerically outpaces Nvidia's, though from a base roughly eleven times smaller. The valuation gap between the two names frames the risk clearly: AMD trades at a trailing P/E of 182 versus Nvidia's 34. The market is pricing AMD's growth trajectory at a multiple that assumes continued share gains and near-flawless execution — a dangerous position if Nvidia's August 26 call delivers upside that resets the competitive narrative. Conversely, any guidance miss from Nvidia that hints at customer diversification toward AMD would compress NVDA's multiple while potentially re-rating AMD higher, at least temporarily.
The hedge fund positioning data adds another layer. Third Point fully exited positions in Nvidia, Meta, KLA, and Lam Research during Q2 2026. That is not a routine rebalance — it is a complete exit from a concentrated AI hardware basket by a fund that had been constructively positioned. Meanwhile, Appaloosa's David Tepper added 322,500 shares to his TSMC stake, implicitly expressing confidence in the foundry layer of the AI trade rather than the fabless layer. The divergence between these institutional moves tells traders that smart money is not monolithic on AI hardware into this earnings cycle.
What Traders Watch on August 26
The headline EPS and revenue figures will move the stock in the first minute after the print. What sustains or reverses that move is the guidance commentary, and specifically two line items that carry more information than the quarterly numbers themselves. First: does management reaffirm, raise, or quietly narrow the $145 billion supply commitment? Any softening of that figure suggests demand signals from hyperscalers are weakening at the margin. Second — and this is the sleeper issue — does the earnings call clarify Nvidia's position on the reported backtrack from its $250 billion OpenAI data center guarantee? If Nvidia is pulling back from that commitment, the strategic logic of its $30 billion OpenAI investment and $10 billion Anthropic stake becomes harder to defend at current valuations.
Export control risk is not hypothetical noise. Chinese AI firms are routing around U.S. restrictions through Southeast Asian data centers, and small batches of H200 chips have already reached ByteDance and Tencent — approximately 10,000 processors each. New U.S. tariffs of 10% to 12.5% on imports from 60 countries add cost friction to a globally distributed supply chain that Nvidia cannot fully insulate. If the Q2 call produces any commentary on tariff impact to margins or on tightened export enforcement reducing addressable market, that is the line traders need to hear above everything else in the script.
Prediction markets are currently pricing a 44% probability of NVDA closing August at $232 and only 49.5% odds of finishing the month above $220. With the stock at $219.74 and up just 17.96% year-to-date — underperforming the expectations embedded in its AI narrative — the setup is asymmetric but not cleanly directional. A clean beat-and-raise to $100 billion or above for Q3 guidance could reclaim the 52-week high of $236.26 within sessions. A miss to $89 billion or softer sequential guidance puts the $200 level back in play fast. Mark August 26 post-market. There is no more important data point in tech this quarter.
Vanguard hauled in $5.96B Tuesday while Invesco shed $4.61B. The rotation into T-bill ETFs and out of credit reveals exactly what the jobs report did to rate expectations.
Roundhill's DRAM ETF tops $23B in 2026's breakout launch. XRP ETFs pulled $150M in August while spot Bitcoin ETFs shed $236.5M in a single September session.
WTI crude surges toward $94.40 on Hormuz deal talks, reigniting Fed rate-hike fears after Friday's 162,000 jobs print. What energy traders must watch today.