The Nvidia Paradox: Why a Veteran Analyst's Bullish Battle Cry Collides With $260 Billion in Vanishing Market Value

On the evening of February 25, 2026, Nvidia delivered what should have been a victory lap. Record revenue. Blowout guidance. A CEO proclaiming the dawn of "agentic AI." By any conventional measure, the numbers were staggering. But by the close of trading the following day, $260 billion in shareholder wealth had evaporated — the sharpest single-day decline for the AI chipmaker in nearly two years [1][2].

Into this chaos stepped Dan Ives, the veteran tech analyst at Wedbush Securities, with a message that startled even hardened Wall Street observers: buy more. His reasoning, and the fierce counter-arguments against it, reveal a fault line running through the most important stock in the world.

The Numbers That Should Have Won the Day

Nvidia's fourth-quarter results for fiscal 2026, reported on February 25, were objectively extraordinary. Total revenue hit $68.1 billion, up 73% year-over-year and 20% sequentially [3]. Data-center revenue — the engine of the AI boom — surged 75% to $62.3 billion, driven by insatiable demand for the company's Blackwell GPU architecture [4]. Full-year revenue for fiscal 2026 reached $215.9 billion, a 65% increase from the prior year [3].

Earnings per share came in at $1.62 on a non-GAAP basis. Gross margins held at a commanding 75.2% [3]. And then came the guidance: $78 billion in expected first-quarter revenue for fiscal 2027, plus or minus 2% — a figure that sailed past consensus estimates and implied continued sequential acceleration [3][5].

CEO Jensen Huang, characteristically ebullient on the earnings call, declared that "the world is now awakened to the agentic AI inflection" and said Nvidia expects sequential revenue growth throughout calendar 2026, exceeding the $500 billion Blackwell and Rubin revenue opportunity the company had previously outlined [4][6]. He described Blackwell demand as continuing to be "off the charts" [7].

By almost every traditional metric, this was a company executing at the highest level in corporate history.

Dan Ives: "Michael Jordan-Like Numbers"

Dan Ives did not hedge. The Wedbush managing director and global head of tech research, who has built his reputation on aggressive but often prescient tech calls, called Nvidia's quarter a "superstar performance" and compared its financial output to "Michael Jordan-like numbers" — a reference to the kind of dominance that defies the usual rules of gravity [8][9].

Ives's core argument is that the market is committing a fundamental error: applying the "law of large numbers" critique to a company that is still in the early innings of a generational platform shift. In his view, the usual expectation that growth must decelerate as a company scales simply does not apply to Nvidia — not yet, and not for the foreseeable future [8].

Following the earnings report, Ives raised his price target on Nvidia to $300, with a base case of $250 by the end of 2026 [9][10]. He argued on CNBC that "the Street's estimates significantly underestimate demand" and that Nvidia's next phase extends well beyond the current data-center build-out into physical AI, robotics, autonomous systems, and enterprise deployment — verticals that will sustain demand for 18 to 36 months at minimum [8][11].

"This is not just more AI," Ives told Bloomberg. He envisions a market that eventually supports five, seven, or even ten important chipmakers, but with Nvidia remaining the unquestioned leader throughout the transition [8].

$260 Billion Says Otherwise

The market's response was a brutal repudiation — at least in the short term. Nvidia shares dropped more than 5.5% on February 26, erasing approximately $260 billion in market capitalization and dragging the S&P 500 and Nasdaq-100 lower with it [1][2][12].

The paradox was not lost on observers. "In reality, they were a huge disappointment," CNN reported, capturing the dissonance between record financial performance and investor sentiment [13]. Mike O'Rourke, chief market strategist at JonesTrading, described the situation bluntly: investors see "trouble in paradise" [13].

Several specific concerns drove the sell-off.

The Capex Cliff Question

The most persistent anxiety surrounding Nvidia is the sustainability of hyperscaler capital expenditure. Amazon, Microsoft, Google, and Meta — Nvidia's most critical customers — are each spending in excess of $100 billion individually on AI infrastructure in 2026, with total Big Five hyperscaler capex surging to $602 billion, up 36% year-over-year, with 75% tied to AI [14].

But growth rates are decelerating. Goldman Sachs expects AI spending to grow 62% in 2026, down from 73% in the prior period [14]. Consensus forecasts call for Nvidia's own sales growth to slow to 58% in 2026 and then 28% in 2027 [14]. The question haunting investors is simple: what happens when the spending cycle peaks?

Jensen Huang pushed back forcefully on the earnings call, citing a path to $600 billion or more in annual AI capex and insisting that current spending levels are "just the start of something far bigger" [15]. But the market's collective response suggested that proclamations alone are no longer sufficient.

The $95 Billion Commitment Problem

Perhaps the most alarming disclosure buried in Nvidia's earnings report was the surge in "purchase obligations" — commitments to suppliers, primarily Taiwan Semiconductor Manufacturing Company (TSMC), for chip fabrication. These obligations ballooned to $95 billion, up from $16 billion a year earlier [13].

These are non-cancellable orders placed well before demand is fully known. In effect, Nvidia has been forced to make enormous bets on future demand to secure manufacturing capacity. If AI spending decelerates faster than anticipated, Nvidia would be left holding commitments vastly in excess of actual orders — a scenario that would crush margins and earnings [13].

The Customer Concentration Trap

Nvidia derives approximately 91% of its revenue from data centers, and roughly 85% from just six customers, with the top four — Microsoft, Amazon, Google, and Meta — accounting for nearly 60% of total sales [2][16]. This creates what analysts call "binary risk": any pullback from a single hyperscaler would cascade through Nvidia's financial results with outsized impact.

Adding to the unease, reports surfaced around the earnings period that Nvidia's anticipated $100 billion deal with OpenAI had stalled, removing one anticipated catalyst from the near-term narrative [13].

The Custom Silicon Threat

While Nvidia's dominance in AI training workloads remains unchallenged for now, every major hyperscaler is investing in proprietary AI accelerators: Google's TPUs, Amazon's Trainium chips, Microsoft's Maia processors, and Meta's MTIA [16]. These custom chips are not yet direct replacements for Nvidia's GPUs in training, but they are creating viable alternatives for inference workloads — the faster-growing segment of AI compute — where Nvidia's CUDA software moat is thinner [16].

Over time, these efforts could gradually erode Nvidia's market share at the margins, even if the total addressable market continues to expand.

The Infrastructure-to-Revenue Gap

There is a deeper structural concern that transcends any single quarter. Hyperscalers committed nearly $400 billion in capital expenditure in 2025, while enterprise AI generated approximately $100 billion in actual revenue [14]. This four-to-one ratio between infrastructure spending and the revenue it generates raises a fundamental question about the sustainability of the current build-out.

If AI applications do not begin monetizing at a pace commensurate with the infrastructure being deployed, the hyperscalers will eventually rationalize spending — and Nvidia, sitting at the top of the supply chain, would feel the effects first.

The Bull Case Holds — For Now

Despite all of this, the overwhelming consensus on Wall Street remains bullish. Of 66 analysts covering Nvidia, 61 rate it a buy or strong buy [5]. The company trades at a forward price-to-earnings ratio that, while elevated in absolute terms, has actually compressed relative to its growth rate — leading some analysts to argue it is undervalued [17].

Tom Graff, chief investment officer at Facet, offered a more measured assessment, predicting "a bumpy ride" for at least the next couple of quarters but noting that the long-term trajectory of AI adoption remains intact [5].

Jim Cramer, for his part, told viewers not to panic over the post-earnings sell-off, framing it as a buying opportunity rather than a structural breakdown [18].

And Ives's argument — that Nvidia's expansion into robotics, autonomous vehicles, physical AI, and sovereign AI infrastructure creates demand vectors that most models do not yet capture — has genuine analytical merit. If even a fraction of these markets develops as projected, Nvidia's current revenue base could be just a waypoint.

What the Nvidia Paradox Tells Us

The February 2026 earnings cycle crystallized something important about Nvidia and about the AI trade more broadly. The company is simultaneously the most dominant franchise in technology and the stock most vulnerable to a change in narrative.

Nvidia is not being punished for poor performance. It is being punished for the possibility that its extraordinary performance might not accelerate forever. That is a very different kind of risk — and one that no earnings beat, however spectacular, can fully resolve.

Dan Ives is betting that the market is wrong, that the doubters are applying old frameworks to a new reality, and that Nvidia's best years are still ahead. The $260 billion that vanished in a single trading session is betting that even Michael Jordan eventually retired.

Both sides may ultimately be right — just on different timescales. And for the investors caught in between, that is precisely the problem.