Nvidia CEO Just Crowned the “Next Trillion-Dollar” Chip Stock and It Went Up 33%

• Marvell Technology stock jumped 33% in one trading session after Jensen Huang’s endorsement, adding $56 billion in market capitalization.
• Nvidia’s top three customers now account for 64% of accounts receivable, up from 56% last quarter and 33% two years ago.
• Huang’s appearance at Computex in Taipei emphasized Marvell’s role in AI data center networking, following Nvidia’s $2 billion equity investment in the chipmaker.

• 33% Single-day stock price gain for Marvell following Huang’s remarks at Computex.
• 64% Share of Nvidia accounts receivable held by top three customers, up from 33% in 2020.
• $56B Market capitalization added to Marvell in one trading session after the endorsement.

On June 2, Nvidia Chief Executive Officer Jensen Huang made an unannounced appearance at Computex in Taipei during Marvell Technology CEO Matt Murphy’s keynote presentation.

Spending roughly ten minutes on stage, Huang publicly called Marvell the “next trillion-dollar company,” citing the firm’s critical role in building networking and connectivity infrastructure for large-scale AI data center deployments.

The remark catalyzed an extraordinary market reaction: Marvell shares surged 33 percent in a single session, the stock’s largest one-day gain on record, and the move added approximately $56 billion in market value, pushing Marvell’s total capitalization above $250 billion for the first time.

The endorsement was not made in a vacuum. Nvidia disclosed a roughly $2 billion equity investment in Marvell earlier, tying the latter’s custom accelerators and optical networking products directly into Nvidia’s stated “AI factory” architecture.

Huang’s public backing thus functioned as both a validation of that strategic partnership and a signal to the broader market that Nvidia views Marvell as central to scaling AI workloads across distributed data centers, where thousands of linked processors must exchange data at unprecedented speeds.

Huang Spotlights Marvell’s Networking Bottleneck as AI Scaling Challenge

The substance of Huang’s remarks centered on a specific technical constraint: as AI clusters grow, raw compute power and memory bandwidth cease to be the limiting factors. Instead, the ability to move data between processor nodes becomes the critical architecture problem.

Marvell manufactures the switches, optical interconnects, and custom silicon that address this bottleneck in hyperscale data centers, products that now generate the bulk of its revenue.

Market analysts bullish on Marvell point to this structural argument with conviction. They contend that as hyperscalers continue building out AI training and inference capacity, the demand for premium connectivity solutions will remain robust. Marvell’s installed base in major cloud providers and the technical complexity of its products create defensibility against commoditization.

The firm competes primarily with Broadcom in networking silicon, though each vendor has differentiated product lines.

However, skeptics raise a countervailing concern: Marvell trades at a steep valuation multiple relative to historical and peer benchmarks, leaving little room for execution missteps or demand disappointment.

Michael Burry’s Nvidia Put Trade Flags Acute Customer Concentration Risk

While Huang’s words were driving Marvell higher, a contrasting view gained traction in institutional circles. Michael Burry, the investor whose early recognition of mortgage-backed security risks formed the basis of “The Big Short,” signaled significant skepticism about Nvidia’s fundamental health.

Through his firm, Scion Asset Management, Burry accumulated put options on one million Nvidia shares, a direct bet that the stock will decline, and made public his thesis on the company’s hidden fragility.

Burry’s core argument centers on customer concentration. In Nvidia’s most recent quarterly filing, the top three customers accounted for 64 percent of the company’s accounts receivable.

This represents a sharp deterioration from the prior quarter, when that figure stood at 56 percent, and represents a dramatic shift from 2020, when the top three customers represented only 33 percent of receivables. The trend reflects the concentration of AI infrastructure spending among a handful of hyperscalers, Meta, Google, Amazon, and Microsoft, who collectively dominate cloud AI investment.

Burry’s concern is straightforward: if demand from those three customers weakens materially or if they shift to alternative chip vendors, Nvidia faces a cliff scenario where the apparent stability of its growth narrative collapses. The single-quarter jump of eight percentage points in concentration underscores how rapid and severe this shift has been.

Unlike a diversified chip manufacturer serving dozens of end-market segments, Nvidia has become structurally dependent on a handful of buyers, each capable of deferring spending independently.

Burry Characterizes Current AI Spending as Temporary “Tokenmaxxing” Bubble

Beyond concentration risk, Burry articulated a more fundamental thesis about the nature of current AI infrastructure spending.

He described a significant portion of current hyperscaler expenditures as driven by a temporary benchmarking phase he termed a “tokenmaxxing bubble”, a period of elevated spending designed to demonstrate capability and competitive positioning in AI, rather than to generate revenue-driving applications.

In Burry’s framing, demand that appears permanent today may prove ephemeral once hyperscalers shift from experimental AI buildout to production-grade deployment of monetized services.

This view stands in direct tension with the bullish AI infrastructure narrative that has driven Nvidia’s valuation to near $3 trillion and has justified rich multiples across the semiconductor supply chain.

If Burry is correct that a significant component of current demand is speculative rather than fundamental, the subsequent normalization of spending could trigger sharp valuation compression not only in Nvidia but across networking, memory, and accelerator suppliers like Marvell.

The question thus becomes whether institutional investors treat Huang’s Marvell endorsement as a durable signal about long-term architecture needs, or as evidence of enthusiasm at the peak of a spending cycle.

Market Tests Conflicting Narratives on AI Demand Permanence

The contrast between Huang’s bullish remarks and Burry’s bearish positioning reveals a critical unresolved debate within the institutional investment community. Nvidia’s own guidance and analyst consensus continue to reflect confidence in sustained AI infrastructure spending through 2025 and beyond.

Yet the speed and magnitude of customer concentration at Nvidia suggests that the current demand cycle is not evenly distributed, it is driven by a handful of firms making extraordinary capital allocation decisions under competitive pressure.

The endorsement of Marvell may be read two ways. First, as a signal that Huang and Nvidia’s leadership genuinely believe networking and connectivity represent a durable, decades-long architecture shift in computing, one that justifies continued heavy investment from Marvell.

Second, as a sign that Nvidia is actively working to diversify its ecosystem and reduce its own dependence on a narrower set of customers, by elevating suppliers like Marvell that can absorb demand across multiple hyperscalers and geographic regions.

Institutional investors face a methodological challenge: Huang’s credibility in predicting chip market inflection points is substantial, but so is the historical record of infrastructure supercycles that peak and correct sharply.

The critical data point to monitor will arrive in Nvidia’s next quarterly earnings report and 10-Q filing, where accounts receivable and customer concentration metrics will either continue to worsen or begin to stabilize. If the concentration continues to exceed 60 percent in the next quarter, Burry’s framing will gain institutional weight, and Marvell’s 33 percent gain may prove ephemeral. Convers

Original reporting: beincrypto.com