Author: 137Labs
On February 25, 2026, NVIDIA (NVDA), the global leader in AI chips, released its fourth quarter and full-year financial results for fiscal year 2026 (ended January 25, 2026): revenue, profits, and data center income all significantly exceeded expectations, while the guidance for the next quarter was further raised. According to the traditional logic that “earnings drive stock prices,” such earnings results typically signal a high likelihood of upward price movement.
The market, however, gave the opposite answer. The day after the earnings report, NVDA's stock price fell by approximately 5.46%, leading to widespread reports of “about $260 billion in market value erased in a single day.” This sharp disconnect between strong fundamentals and weak stock performance stems not from the authenticity of the results, but from a shift in capital markets’ weighting—from quarterly profits toward growth duration, capital expenditure slope, and structural risks.
First, lock down the earnings report: How strong is it?
According to NVIDIA's official disclosures, the key data for Q4 and the full fiscal year 2026 are as follows:
Q4 Revenue: $68.127 billion, +73% year-over-year, +20% quarter-over-quarter
Q4 data center revenue: $62.3 billion, up 75% year-over-year and 22% quarter-over-quarter, continuing to set new records
·Q4 GAAP net profit: $42.96 billion; Non-GAAP net profit: $39.55 billion
· Annual revenue: $215.938 billion, +65% year-over-year
· Full-year GAAP net profit: $120.067 billion
· Guidance for next quarter (Fiscal Q1 2027): Revenue of approximately $78 billion (±2%)
This data implies two things: first, demand for AI infrastructure remains in a phase of strong expansion; second, NVIDIA’s revenue structure is further concentrating on the “data center single engine.”
II. Strengths Are Becoming Single Points of Risk: Excessive Reliance on Data Centers
The standout feature of the earnings report is also the market’s most sensitive point: Q4 data center revenue reached $62.3 billion out of total revenue of $68.1 billion, accounting for approximately 91.5%. This means NVIDIA has almost entirely bet its growth on the “AI capital expenditure cycle”—the more cloud providers, sovereign nations, and large enterprises invest in computing power, the more NVIDIA resembles a high-growth machine; but should capital spending shift from expansion to contraction, the resulting volatility will be amplified accordingly.
Meanwhile, non-data center businesses, even if growing, struggle to provide effective hedging. Businesses such as automotive, gaming, and professional visualization are not on the same scale as data centers. For example, automotive revenue in a single quarter is approximately $604 million—far too small to offset cyclical fluctuations in the data center business. This structure is viewed as “highly focused efficiency” during bull markets but quickly transforms into a discount for “single-engine dependency” at emotional turning points.
Three: Increased customer concentration—control is in the hands of a few
The market often summarizes NVIDIA’s customer structure with the phrase “the five major cloud providers account for more than half of revenue.” In fiscal year 2026, NVIDIA’s sales concentration increased, noting that two customers together accounted for 36% of sales. The conclusion is straightforward—NVIDIA’s supergrowth is deeply tied to a small number of ultra-large customers.
This binding has a double-edged effect:
· Bull phase: The faster leading clients expand, the more NVIDIA can "tax."
· Downturn phase: Once major clients slow their capital expenditures, NVIDIA's orders and valuation will both come under pressure;
A more subtle risk lies in the shift in bargaining power: as customers systematically support second suppliers or develop in-house alternatives, NVIDIA’s “monopoly premium” will be compressed into a “leadership premium.”
The post-earnings market decline largely reflects an early discounting of the combined risks of "growth concentration + shifting bargaining power."
IV. Why Did "Better-Than-Expected" Become a Negative? The Pricing Logic Has Shifted from Quarterly to Duration
NVIDIA has consistently exceeded expectations for multiple quarters, causing the very notion of "beating expectations" to gradually lose its marginal surprise. Ahead of the earnings report, capital had already fully priced in a "strong report" through positioning and derivative structures, resulting in a typical trading outcome: no matter how strong the earnings are, if they fail to introduce new catalysts beyond the existing narrative, they are prone to trigger profit-taking.
This type of price movement often manifests as “good news priced in.” When the market is anticipating growth over a 2027 horizon or even longer, what financial reports most need to address is not whether performance can continue to “blow out” in the current quarter, but rather how long growth can be sustained, through what structure, and under what competitive conditions. Without greater long-term certainty, an unusual combination of strong fundamentals but weak stock prices may emerge.
V. Is the AI bubble a false proposition? More like a reassessment of capital expenditure and credit
The "AI bubble" is often misinterpreted as "AI has no value." A more accurate reflection of the real disagreement is: AI's value is undeniable, but the mismatch between investment and return timing is being seriously priced in.
Cloud providers' AI capital expenditures continue to rise significantly, with massive investments, yet commercial returns are still in the early growth phase. Amid high interest rates or profit pressures, the market naturally asks: When will such substantial compute investments translate into sustainable profits? If short-term returns remain negative—where spending continues without profit—the valuation baseline for upstream compute suppliers could be reassessed if the pace of capital expenditure slows.
This is no stranger to the cryptocurrency industry’s cycle: infrastructure expansion often precedes application adoption. When “supply expansion” outpaces “demand realization,” prices and valuations become highly sensitive to shifts in sentiment. AI is now at a similar stage, except this time the “ledger” isn’t on-chain—it’s in the financial reports of cloud providers and semiconductor leaders.
Six: The Real Competitive Threat—It’s Not That “Someone Can Make GPUs,” But That “Customers Don’t Want to Buy from Just One Supplier”
For a long time, NVIDIA built its moat through GPU leadership, the CUDA ecosystem, and integrated system solutions. However, the key shift in the competitive landscape is not due to any single company’s breakthrough, but rather a structural change on the customer side—adopting a second supplier, developing in-house chips, and replacing single-GPU purchases with full system solutions.
1) AMD × Meta: Second Supplier Strategy Becomes Institutionalized
Meta’s long-term, high-value partnership with AMD is not primarily aimed at immediately shifting market share, but rather at sending a signal: major customers are using committed orders to support alternative solutions and reduce reliance on a single supplier. The direct consequence of this strategy is a marginal decline in NVIDIA’s pricing power in future negotiations, thereby compressing its valuation premium.
2) The Arrival of the Inference Era: The Compute Race Shifts from "Training" to "Cost and Latency"
The focus of the AI industry is gradually shifting from cost-unconscious training to cost-sensitive inference. Inference emphasizes throughput, latency, energy consumption, and cost per unit, creating opportunities for more specialized new architectural players. NVIDIA is addressing its weaknesses by integrating inference-related technologies and teams—for example, through technology licensing and personnel integration with inference chip company Groq—demonstrating that competition in the inference era has evolved from “chip performance” to a full-stack system efficiency battle.
Seven: NVIDIA is building its second curve: from cloud computing power to an operating system for the physical world.
Viewing NVIDIA merely as a company that sells GPUs underestimates its strategic depth. During earnings cycles, NVIDIA continues to advance platform initiatives in “physical AI” domains such as autonomous driving, robotics, and industrial simulation, while introducing open-source capabilities for autonomous driving inference and safety validation (e.g., Alpamayo). While this line of business contributes limited short-term value, it signals a broader direction: elevating NVIDIA from a “seller of shovels” to a provider of operating system-level infrastructure, locking customers into platforms and ecosystems rather than just hardware purchases.
Once this platformization succeeds, NVIDIA’s growth duration will no longer be solely determined by cloud providers’ capital expenditures, but will increasingly stem from longer-cycle demands such as industrial digitalization, industrial robotics, and autonomous driving. However, before this second growth curve achieves true scale, the market will still primarily value NVIDIA using the framework of a “data center single-engine + capex cycle asset.”
Eight: Key Variables for 2026—Stock Price Is Determined by Three Curves, Not Just a Profit and Loss Statement
The core determinant of NVIDIA's valuation center in 2026 is not whether it can continue to grow, but how long the growth can be sustained and through what structure. The market will primarily focus on three verifiable trends:
1) Cloud provider capital expenditure trend: Continuing to accelerate, or showing marginal slowdown?
2) Revenue Structure and Systemic Penetration: Can the transition from “selling GPUs” to “selling comprehensive system solutions (network interconnection, software stack, platform tools)” continuously enhance customer retention and average revenue per customer?
3) Speed of second supplier and in-house development adoption: The faster alternative solutions move from pilot to large-scale procurement, the more easily NVIDIA’s premium pricing space is compressed.
Conclusion: The earnings report proves that the hash power myth continues, but pricing has entered the "duration test."
This earnings report confirms that the AI infrastructure boom continues, and NVIDIA remains the most powerful cash-generating machine for computing power. However, the stock price decline serves as a reminder to the market: when “explosive growth” becomes the norm, pricing logic has shifted from growth rate to sustainability, from profitability to growth duration, and from monopoly premiums to competitive dynamics.
Post-earnings adjustments do not necessarily signal a fundamental reversal, but rather a shift in valuation focus. NVIDIA remains strong, but the real test lies in how long growth can be sustained and whether its structure can become more stable.
This answer will determine the valuation boundary for NVIDIA in 2026 and influence the direction of risk appetite for AI assets.