The market is becoming increasingly picky about AI: NVIDIA's performance exceeding expectations this year is no longer enough; the key is to look at the revenue visibility for 2027

NVIDIA's upcoming fourth-quarter earnings report is expected to exceed expectations once again, but investors' focus has already shifted to the more distant future. The current stock price has largely absorbed the growth potential for 2026, and whether the stock can continue to rise depends on the company's visibility into 2027 revenues.

According to the Wind Trading Desk, Goldman Sachs analyst James Schneider in a recent report predicts that NVIDIA's fourth-quarter revenue will exceed market expectations by about $2 billion, with the firm's first-quarter revenue forecast 8% higher than the Wall Street consensus. However, he also pointed out that the $2 billion beat may only be "in line with expectations," and what truly drives the stock price is guidance for the next 12-18 months. Goldman Sachs maintains a buy rating and a target price of $250, representing a 43.5% upside from the current $174.19.

Specifically, Goldman Sachs expects fourth-quarter revenue of $67.34 billion (market expectation: $65.64 billion), adjusted earnings per share of $1.59 (market expectation: $1.52); first-quarter revenue of $76.84 billion (market expectation: $71.15 billion), adjusted earnings per share of $1.80 (market expectation: $1.65). The data center business remains the core engine, with Goldman Sachs expecting fourth-quarter revenue for this segment to be $61.3 billion, rising to $71.1 billion in the first quarter.

More striking is the valuation difference. Goldman Sachs' earnings per share forecasts for fiscal years 2026 and 2027 are 17% and 29% higher than market consensus, respectively. This suggests either that Goldman Sachs is far more optimistic about growth than the market, or that there is a significant pricing error in the market.

$500 billion forecast requires a clearer roadmap

NVIDIA previously set a long-term goal for data center business revenue to reach $500 billion. Investors now want to know: when will this number be achieved, what is the customer composition, and how much will computing and networking businesses each account for.

Starting from this already significantly above-market expectation baseline, Goldman Sachs believes that any positive statement from NVIDIA regarding visibility into 2027 revenues could serve as a catalyst for the stock price. According to Goldman Sachs' model, data center revenue for fiscal year 2027 is expected to reach $357.3 billion, 16% higher than market expectations; further increasing to $483.9 billion in fiscal year 2028, 22% above market expectations.

The pace of product transitions is also a key variable. Goldman Sachs expects Rubin GPUs to start shipping in the third quarter, with significant ramp-up in the fourth quarter and beyond. According to the model, by the first quarter of fiscal year 2027, Rubin will account for a major share of revenue, while Blackwell's share will decline rapidly. This rapid product iteration is both NVIDIA's technological moat and a test of supply chain execution capabilities.

When will demand from non-traditional customers materialize?

OpenAI is expected to begin large-scale deployment in the second half of 2026, aiming to build approximately 26GW of computing power within 4-5 years. Although the procurement volume for 2026 is still small relative to this long-term goal, any initial signs of execution are worth noting.

Goldman Sachs points out that OpenAI is working with NVIDIA, Broadcom, and AMD to advance deployment. In addition to OpenAI, Anthropic has raised its revenue expectations for the calendar year 2026 by 20%, and sovereign AI deployment activities remain strong The extent to which the incremental demand from these non-hyperscaler cloud customers will offset the fluctuations of traditional customers is a source of uncertainty for the revenue forecast in 2027.

Quantitative data on the increased capital expenditures of U.S. hyperscaler cloud vendors for 2026 and 2027, specific details on the demand from non-hyperscaler customers in 2026, and the performance of large language models trained on NVIDIA's latest chips—these three types of information will be released successively in the first half of the year, forming a catalyst path for stock prices. The GTC conference in March will be an important observation window.

Intensifying Competition but CUDA Remains a Barrier

Google's TPU v7, AMD's MI455X, and Microsoft's Maia 200 are expected to be closer to NVIDIA's products in terms of raw computing power performance. The efforts of hyperscaler cloud vendors to develop their own ASICs are increasing, posing a potential threat to NVIDIA's market share.

Goldman Sachs expects NVIDIA to emphasize the competitive advantages of the CUDA ecosystem. This software platform has developed a network effect among developers over many years. Recent transactions between NVIDIA and Groq and their impact on inference costs are also worth noting—the growth rate of the inference market may exceed that of the training market, and the competitive landscape in this segment has not yet fully formed.

The Chinese market may open up space for revenue contributions before 2027, with the specific scale and timing of revenue contributions needing further disclosure from management.

Valuation Bets on Structural Demand

Goldman Sachs' target price of $250 is based on a 30x price-to-earnings ratio applied to a normalized earnings per share of $8.25. This means that even if the growth rate of AI infrastructure spending slows, NVIDIA can still maintain high profitability.

The current stock price corresponds to a price-to-earnings ratio of about 20x for the fiscal year 2027 (according to Goldman Sachs' forecast), and about 14x for the fiscal year 2028. If one accepts Goldman Sachs' growth forecast, the valuation is not considered high. However, the issue lies in the probability of achieving Goldman Sachs' forecast itself—an EPS adjustment potential of 29% requires demand to continue exceeding expectations or gross margins to remain at a high level of around 75%.

Major risks include a slowdown in AI infrastructure spending, market share erosion by ASICs and AMD, gross margin declines due to competition, and supply chain constraints. Goldman Sachs' model shows that operating profit margins will stabilize in the range of 67-69% for the fiscal years 2027-2029, which requires the company to maintain cost control and pricing power while expanding its scale