Nvidia Corp. (NVDA) reports fiscal first-quarter 2027 earnings Wednesday after the close, and the setup is about as tense as it gets for the AI trade.
The latest Bank of America Global Fund Manager Survey found "long global semiconductors" is the most crowded trade in the survey's history, with 73% of managers calling it that. Meanwhile, the 30-year U.S. Treasury yield just hit a 19-year high of 5.2%. Options markets are pricing an 8.65% swing in Nvidia stock after earnings, in either direction.
What Nvidia says Wednesday night, and how it guides for the quarter ahead, will set the tone for the rest of 2026 across semiconductors, AI infrastructure, and the broader market.
What Wall Street Expects
According to MarketDash data, analysts expect Nvidia to report earnings per share of $1.76 on revenue of $78.67 billion. That would be a 117% jump in earnings from the same quarter last year, when Nvidia reported EPS of $0.81 on revenue of $44.06 billion — and a 79% revenue increase.
Bank of America semiconductor analyst Vivek Arya, who has a Buy rating and $320 price target on Nvidia, calls it his "top pick" heading into the print. He notes that Nvidia's last 10 beats have averaged 7-8% above guidance, which would imply sales of $83-84 billion.
Arya highlights five key debates that will decide the print and the AI trade's direction.
1. Nvidia's Cash Returns
Nvidia's "large existing positioning often acts as a headwind," Arya says. The stock now makes up 8.3% of the S&P 500 — above the peak weightings of Apple (AAPL) in June 2023 and Microsoft (MSFT) in November 2023. Active fund manager ownership sits near 78%.
The problem is the marginal buyer. Apple and Microsoft expanded their shareholder bases by returning capital aggressively to dividend and income-oriented funds. Nvidia hasn't done that. Only 47% of its free cash flow from 2022 to 2025 went to dividends and buybacks, versus a peer average of 80%. Its dividend yield is 0.02%, compared with a peer average of 0.89%. And only 16% of equity income funds own Nvidia, while peers average 32%.
Nvidia's ecosystem investments in OpenAI, Anthropic, and other tech partners have been "unfairly, in our view, characterized as circular/vendor financing," Arya says. He sees a potential capital-return announcement as a second-half catalyst that could close the valuation gap and neutralize that narrative.
2. Vera Rubin Ramp Timing
The successor to Nvidia's Blackwell architecture, Vera Rubin, is expected to ramp in the second half of 2026. Arya notes that Vera Rubin uses the same Oberon rack architecture as Blackwell and Blackwell Ultra, which should enable a "generally easy product transition in 2026-27 without much GM impact." Vera Rubin Ultra, arriving in the second half of 2027, brings a new Kyber rack architecture and higher HBM content per system.
The competitive timeline matters: Advanced Micro Devices (AMD) is expected to launch its MI400 "Helios" platform in the same window.
3. How Long Can Nvidia Sustain Its Margins?
Nvidia's gross margin sits near 75%. Consensus models are converging toward 74% over time as HBM costs rise and Rubin Ultra introduces architectural changes. But Arya argues that consensus understates margin durability. Nvidia has held margins through multiple product cycles, including a modest setback during the initial Blackwell ramp in early 2025. Rubin's reuse of the Oberon architecture should support continuity through 2026 and 2027.
HBM pricing has more than doubled this cycle as supply lags AI memory demand — a cost headwind that Nvidia has absorbed so far through pricing power and a mix shift toward higher-end systems.
4. The $1 Trillion Forecast
Nvidia previously guided to a $1 trillion revenue stack for calendar years 2025 through 2027. The next update will include contributions from LPU racks, the Vera CPU, and Vera Rubin Ultra — components not embedded in the original number, according to Arya.
BofA's own estimate for the AI data center addressable market now sits at $1.7 trillion by 2030, up from $1.4 trillion in February and $823 billion in June 2025. Arya expects Nvidia to "maintain 70% share of overall $1.7Tn+ AI TAM by CY30," with the company retaining roughly 65-70% of the AI accelerator market through the end of the decade.
Hyperscaler capex from Microsoft, Amazon (AMZN), Alphabet (GOOGL), and Meta (META) is now running at 95-100% of operating cash flow, compared with a historical 35-50%.
5. Competitive Landscape
Concerns about Alphabet's TPUs, custom ASICs, and agentic CPUs have weighed on sentiment. But Arya says Nvidia's "product/customer breadth, software/developer support, multi-cloud availability create an unparalleled standardized infrastructure" that other merchant or custom chip competitors would struggle to overtake.
The agentic CPU narrative — that CPU-to-GPU ratios are tilting in favor of CPU — is "misleading," Arya says. Nvidia's own Vera CPU will be "a formidable rival in stand-alone CPU," with news expected at Computex in early June. The CPU/GPU ratio in widely deployed Blackwell and TPU clusters already sits at 1:2, with no meaningful correlation to agentic CPU in standalone clusters. CPU "will be a large but crowded market with x86/ARM options," Arya adds, alongside Intel (INTC) and Arm-based competitors.
What Wall Street Is Saying
Nvidia carries a consensus price target of $285.06 across 34 analysts tracked by MarketDash, implying roughly 30% upside from the $220 area where shares traded Tuesday morning. The Street-high is $360 from Tigress Financial; the Street-low is $215 from Deutsche Bank.
Three analysts updated their views on May 18 ahead of the print. DA Davidson reiterated a Buy with a $300 target. Morgan Stanley's Joseph Moore lifted his target to $285 from $260, maintaining Overweight. Wedbush also weighed in, bringing the three-analyst average to $295. The rating distribution skews heavily toward Buy.
