There is a number sitting inside every serious portfolio manager’s model right now that does not compute on its face. Nvidia trades at under 17 times forward earnings, a multiple that has been cut in half compared with 2025, a period when the company’s expansion was actually slower, and down sharply from above 25 times expected profits as recently as May, Bloomberg reported. Meanwhile, the CEO is telling anyone who will listen that volume is about to double.
That gap is the argument of the week.
Why the Multiple Is Falling
The simple answer is that Wall Street no longer trusts the earnings. Eli Horton, a TCW senior portfolio manager, told Bloomberg that “the stock has de-rated pretty significantly, which suggests a healthy dose of skepticism that the company’s current earnings power is sustainable,” adding that the market appears to be “expecting less than what the consensus is currently estimating.”
The broader semiconductor index has surged this year, driven by outsize gains from names like Micron, Intel, and AMD, while Nvidia ranks among the laggards within the SOX index. The world’s most powerful AI chip company is being lapped by the companies it used to lap. The semiconductor index itself now carries a valuation of about 20 times forward earnings, meaning Nvidia trades at a discount to its own sector.
The Bull Case: A Decade’s Worth of Cheap
Huang told reporters on the sidelines of a King Charles AI summit in Scotland that Nvidia expects to sell twice as many chips next year as it does this year, citing AI investment “across industries, different economies” as the driver. That is not a hedge. It is the most explicit volume commitment the company has made.
Analyst consensus earnings-per-share for fiscal 2028 has moved sharply higher over the past few months, with revision breadth heavily skewed upward. The doubling is being driven by the Vera Rubin platform, which is ramping into full production alongside continued Blackwell deployment and carries a $40 billion per-gigawatt revenue opportunity versus Blackwell’s $25 billion. A company growing earnings that fast at 17 times forward is, by any classical screen, a large-cap growth stock priced like an industrial cyclical.
The Bear Case: The Market Knows Something
The counterargument is that the market is doing its job. Market strategists point to tightening margins and rising competition as core drivers of the valuation discount, with Nvidia’s gross margin expected to step down into the low 70s near term before recovering.
The deeper concern is structural. The CUDA software moat is eroding as hyperscaler customers internalize their AI workloads onto custom silicon stacks where they do not need CUDA. Broadcom CEO Hock Tan has guided that his company’s AI and AI networking revenue opportunity could reach $60 to $90 billion by 2027, an estimate that assumes the hyperscaler ASIC trajectory continues to accelerate. Google’s TPU momentum is already measurable.
Nvidia has never issued guidance this far into the future, and there is no established track record to judge how dependable the 2027 projection will prove. Huang’s doubling forecast is supply-constrained: the CFO framed the fiscal 2028 outlook calling for roughly 70% revenue growth as “a supply-constrained” outlook. Constrained supply means constrained earnings visibility.
What Investors Are Missing
The debate has been framed as GPU demand versus custom ASIC displacement. The overlooked implication is what a sustained Nvidia discount does to every company in its supply chain. Meta still relies primarily on GPU-based AI servers. If hyperscaler capex keeps climbing and one of the largest buyers remains GPU-dependent, Nvidia’s volume promise is far more credible than the multiple implies. The market may be discounting Nvidia’s earnings while simultaneously not discounting the risk that its own bear case is wrong.
Stocks to Watch
NVDA: The central question. At under 17 times forward earnings with roughly 70% fiscal 2028 revenue growth guided, the risk-reward is asymmetric if margins stabilize even modestly above trough.
AVGO: The direct beneficiary of the ASIC displacement thesis. Broadcom co-designs custom silicon for Meta and is the clearest alternative-to-Nvidia trade, with its own $60-to-$90 billion serviceable market guidance.
TSM: Nvidia has partnered with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to mobilize more than $500 billion for AI infrastructure financing over time. TSMC manufactures Nvidia’s advanced Vera Rubin chips and sits at the bottleneck of every scenario, bull or bear.
AMD: AMD’s recent $1 trillion market capitalization milestone reflects its transition from a CPU and GPU supplier into a broader competitor in AI computing infrastructure. It is the second-mover that benefits most if Nvidia’s pricing power softens.
ARM: Arm may be the broadest beneficiary in the sector because it participates in Nvidia’s Vera CPU, hyperscaler processors, and other designs simultaneously, making it the one name that wins regardless of which silicon architecture takes share.
