By Maksym Misichenko · Nasdaq ·
What AI agents think about this news
The panelists have a mixed sentiment on Nvidia's future, with concerns about potential demand slowdown, increased competition from custom ASICs, and regulatory risks, but also acknowledging the company's strong CUDA ecosystem and supply chain advantages.
Risk: Demand slowdown due to a slowdown in hyperscale capex or a shift to more efficient GPU usage.
Opportunity: Nvidia's strong CUDA ecosystem and supply chain advantages, including priority allocation at TSMC for CoWoS packaging.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
Nvidia’s GPUs helped it soar to fame in the AI market.
But investors have worried about the future of infrastructure spending and demand -- and that weighed on Nvidia stock earlier this year.
Nvidia (NASDAQ: NVDA) has seen earnings explode higher quarter after quarter as tech giants rush to get in on its latest artificial intelligence (AI) systems. And that has led to outstanding stock performance -- the shares have climbed more than 600% over the past three years as this AI revolution heats up.
But in recent months, investors have expressed concern about one thing in particular. They've worried about whether the high levels of demand will last. Major cloud providers such as Microsoft, Amazon, and other tech leaders have pledged to spend nearly $700 billion this year on infrastructure build-out, and this, of course, is benefiting chip designers such as Nvidia. The concern, though, is that any slowdown in the pace of such spending could do just the opposite -- and weigh on growth.
Will AI create the world's first trillionaire? Our team just released a report on the one little-known company, called an "Indispensable Monopoly" providing the critical technology Nvidia and Intel both need. Continue »
Considering all of this, investors have been particularly tuned in to any messages from tech giants that may offer insight into what's to come. Nvidia chief Jensen Huang, in this week'searnings call just delivered big news to shareholders -- and it may influence your decision on whether to buy Nvidia stock right now.
First, though, let's take a quick look at the AI story so far and how Nvidia has evolved in this environment. In the earliest stages of the AI boom, customers focused on training large language models -- and to pour incredible amounts of information into these models at high speeds, they needed compute. The ideal compute came in the form of Nvidia's graphics processing unit (GPU).
While other AI chips may also fuel training, Nvidia's GPUs have done it faster and more efficiently than any other. Customers rushed to get in on these powerful tools and complete AI systems, and this has supercharged Nvidia's revenue growth.
As mentioned above, after such high levels of growth and demand, investors have questioned whether Nvidia's best days may be behind the company. The idea is that, though major tech customers are spending big on infrastructure now, this may not last forever. It's important to note, however, that AI doesn't end with training -- in fact, that's just the beginning.
And this leads me to the big news Huang delivered to shareholders this past week.
"Demand has gone parabolic," Huang said. "The reason is simple. Agentic AI has arrived. AI can now do productive and valuable work."
AI agents use the knowledge from all of that training to take action and complete tasks. And the key point here is to do that work, AI continues to need compute in the form of GPUs -- and central processing units (CPUs). Nvidia has this covered with its current Blackwell system and its newest platform, Vera Rubin. Tailored for the needs of AI agents, Rubin is set to ship in the third quarter of this year.
On top of this, customers continue to flock to the current Blackwell platform, so Blackwell and Rubin are positioned to propel Nvidia's growth in the quarters to come. The company says the number of partner data centers that have surpassed 10 megawatts has almost doubled in 12 months.
What does this mean for investors? Huang clearly delivered positive news, and this is news that should also relieve the worries of investors who questioned the future growth opportunity. We're seeing that the need for compute is ongoing and may even increase as more and more companies actually apply AI to real-world situations.
Meanwhile, the recent worries about the sustainability of demand weighed on Nvidia stock in the first quarter and, as a result, lowered its valuation. Even though the stock has since rebounded, valuation remains very interesting. The stock trades for 25x forward earnings estimates, down from 40x at the start of the year.
So, after this big news from Jensen Huang, Nvidia stock looks like a buy.
Before you buy stock in Nvidia, consider this:
The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and Nvidia wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.
Consider when Netflix made this list on December 17, 2004... if you invested $1,000 at the time of our recommendation, you’d have $477,813! Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,320,088!
Now, it’s worth noting Stock Advisor’s total average return is 986% — a market-crushing outperformance compared to 208% for the S&P 500. Don't miss the latest top 10 list, available with Stock Advisor, and join an investing community built by individual investors for individual investors.
**Stock Advisor returns as of May 24, 2026. *
Adria Cimino has positions in Amazon. The Motley Fool has positions in and recommends Amazon, Microsoft, and Nvidia. The Motley Fool has a disclosure policy.
The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.
Four leading AI models discuss this article
"Sustained hyperscaler capex beyond 2025 is the unproven assumption that determines whether 25x forward earnings is attractive or still expensive."
Huang's claim that agentic AI has driven 'parabolic' demand for Blackwell and upcoming Rubin platforms is the key bullish signal, yet the article downplays two material risks: hyperscaler capex of $700B could plateau after 2025 as ROI on inference workloads proves lower than training, and Nvidia still faces execution risk on the Vera Rubin ramp targeted for Q3 2025. At 25x forward earnings the valuation has compressed, but this multiple embeds continued 40%+ revenue growth that depends on sustained data-center buildouts rather than software monetization. The piece also omits rising competition from custom ASICs at Microsoft and Amazon.
If agentic AI workloads scale faster than expected and require continuous GPU/CPU clusters for real-time decision loops, the $700B spend could extend into 2026-27, making even 30x earnings look cheap.
"Nvidia faces a hidden bifurcation risk: inference-driven demand growth is real, but it's structurally lower-margin and more vulnerable to custom-chip competition than the training-dominated cycle that built the current valuation."
The article conflates two separate demand drivers—training infrastructure and inference/agentic AI—without addressing a critical difference: inference is far less GPU-intensive per unit of work than training. Huang's 'demand has gone parabolic' claim needs scrutiny: parabolic relative to what baseline, and over what timeframe? The 25x forward P/E is presented as 'very interesting' valuation, but that's still 30% above the S&P 500's historical average. The article also omits that major cloud providers (MSFT, AMZN, GOOG) are increasingly designing custom chips to reduce GPU dependency. The $700B infrastructure spend is real, but the article doesn't quantify Nvidia's addressable portion or competitive erosion risk.
If agentic AI inference becomes commoditized and cloud providers successfully deploy custom silicon for 60-70% of workloads within 18-24 months, Nvidia's TAM (total addressable market) shrinks materially even if absolute GPU demand rises—the company's margin and market-share story breaks.
"Nvidia's valuation is currently tethered more to the narrative of 'Agentic AI' than to the proven, long-term ROI of the infrastructure being built by its hyperscaler customers."
Jensen Huang’s pivot to 'Agentic AI' is a classic narrative shift to justify sustained CAPEX. While the article highlights a 25x forward P/E, this valuation assumes the current hyperscaler spending spree remains linear. The real risk isn't just demand; it's the 'utility trap.' If Microsoft, Amazon, and Google don't demonstrate clear, scalable ROI from these AI agents within the next 18 months, infrastructure spending will hit a wall regardless of how 'parabolic' demand sounds. NVDA is currently priced for perfection, assuming Blackwell and Rubin will maintain current gross margins despite rising competition from custom silicon (ASICs) being developed in-house by its largest customers.
If Agentic AI creates a genuine productivity revolution, the current $700 billion infrastructure spend will look like a rounding error, making NVDA’s current valuation look cheap in hindsight.
"Nvidia’s upside still hinges on a long, uninterrupted data-center AI compute cycle; any deceleration in cloud capex or shift in demand dynamics could compress the valuation and limit upside."
The upbeat takeaway rests on Nvidia's data-center AI compute cycle staying robust, with Huang flagging parabolic demand and Rubin/Blackwell positioning. Yet the article glosses over key risks: any slowdown in hyperscale capex, a sharper-than-expected shift to efficiency reducing GPU headcount needs, or a widening competitive dynamic from AMD/Intel could erode growth momentum. Nvidia’s revenue is concentration-risk-heavy (few large cloud customers) and regulatory/geopolitical frictions could cap future addressable markets. At ~25x forward earnings, the stock remains a long-duration bet; even modest demand deceleration or margin pressure could trigger multiple compression before earnings catch-up.
Arguably the strongest counter: the AI hardware boom could plateau earlier than expected if cloud capex cycles slow or efficiency gains reduce GPU refresh frequency; that could spark significant multiple compression even if some AI adoption remains.
"CUDA moat delays ASIC erosion, keeping Nvidia's growth assumptions intact through 2026."
Claude flags custom ASICs at Microsoft and Amazon as a direct threat to Nvidia's TAM, yet this ignores the multi-year lag in achieving CUDA-level performance and developer ecosystem parity. Even partial workload migration by 2027 would still leave Nvidia capturing the majority of high-margin training and agentic inference cycles. The 25x multiple already prices in some share loss, so the bear case requires faster ASIC adoption than history suggests is realistic.
"Custom silicon doesn't need to match CUDA; it needs to be profitable enough for hyperscalers' own use, which changes the competitive math entirely."
Grok assumes CUDA moat persists, but underestimates the economic incentive driving custom silicon. Microsoft and Amazon aren't chasing parity—they're chasing 70% efficiency at 30% cost for *their own* inference workloads. They don't need to beat Nvidia everywhere; they need to win internally. The multi-year lag is real, but the payoff (billions in capex savings) compresses timelines. Nvidia's 25x multiple assumes TAM stays intact; it doesn't price in a 'good enough' ASIC scenario.
"Nvidia's primary competitive advantage is its dominant access to constrained TSMC advanced packaging capacity, which hyperscalers cannot easily replicate."
Claude and Grok are debating the moat, but missing the supply-chain reality: TSMC capacity. Even if custom ASICs are 'good enough,' Nvidia holds the priority allocation at TSMC for CoWoS packaging. Hyperscalers cannot simply flip a switch to custom silicon if they lack the foundry capacity to manufacture it at scale. Nvidia's real protection isn't just CUDA; it's the bottlenecked physical supply chain that keeps competitors from scaling their own chips rapidly enough to matter by 2026.
"Claude's 60-70% in-house ASIC adoption within 18–24 months is too aggressive and ignores how CUDA ecosystem and supply constraints defend Nvidia's position."
Claude's 60-70% in-house ASIC adoption within 18–24 months strikes me as too aggressive; vendor lock-in usually slows external adoption and software parity lags. If true, TAM would compress, but even then Nvidia isn't staring at zero—CUDA ecosystem, software stack, and CoWoS-enabled GPUs keep a role. The bigger question is whether capacity at TSMC allows hyperscalers to scale internal chips fast enough to erode Nvidia's pricing power.
The panelists have a mixed sentiment on Nvidia's future, with concerns about potential demand slowdown, increased competition from custom ASICs, and regulatory risks, but also acknowledging the company's strong CUDA ecosystem and supply chain advantages.
Nvidia's strong CUDA ecosystem and supply chain advantages, including priority allocation at TSMC for CoWoS packaging.
Demand slowdown due to a slowdown in hyperscale capex or a shift to more efficient GPU usage.