By Maksym Misichenko · CNBC ·
What AI agents think about this news
The panelists generally agreed that NVDA's valuation is attractive on the surface, but they raised significant concerns about the sustainability of its growth narrative and the risks associated with hyperscaler capex fatigue, in-house chip development, and regulatory headwinds. They also noted that NVDA's stock price may not move even with a beat-and-raise earnings report, and could face a rapid multiple contraction if guidance disappoints.
Risk: Hyperscaler capex fatigue and in-house chip development could gut NVDA's growth narrative and compress its valuation.
Opportunity: NVDA's 24x forward P/E multiple may support its stock price in the near term, given its dominance in data-center GPUs and the potential for strong earnings growth.
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 →
Want to start an Nvidia position but think it's too late? The stock hit an all-time high on Thursday and has quickly dropped 7% since then. We know that Nvidia can swing wildly in the near-term. With earnings coming up on Wednesday evening, we feel that a "beat and raise" quarter is the minimum requirement. It's been the bar for years now, as we wrote in Sunday's earnings preview . Even if that bar is cleared, there are so many moving parts that there's no telling where the stock might trade after the release. We tend to like it when a hot stock cools off a bit ahead of the print. How does that line up with Jim Cramer's column from a couple of weeks ago, when he wrote about the reasons why it might not matter if you are "late" in buying stocks tied to the secular artificial intelligence boom? Listen to Nvidia CEO Jensen Huang talk about AI being the fourth industrial revolution, and the next leg of agentic AI needing 100 times more computing power than currently available, and it might still be early. And, Amazon CEO Andy Jassy? He told Jim earlier this month that investors will reap rewards from all the company's AI spending. It's generally not our style to chase stocks near record highs, nor do we love buying a stock right into an earnings release — even if you're right on the numbers, the price reaction is too hard to game. However, for those without an Nvidia position on the books currently, here is where we stand. During Monday's Morning Meeting, Jim said, "If you wanted to buy some here, I totally endorse it. I just feel that we own it, we don't wanna trade it." Sure, it's hard to think it's anything but late when Nvidia and other stocks have exploded higher in the past three years. But, to never start a position because you are worried that you missed rallies that have already transpired, when the long road ahead looks primed to run, would violate No. 13 on Jim's list of investing rules, "No woulda, shoulda, coulda." If you have not started a position in Nvidia yet, don't let your past inaction prevent you from getting on board. Jim has repeatedly and openly talked about mistakenly exiting Alphabet in March 2025 on concerns about AI and federal litigation. It quickly became apparent that AI was not a problem, and neither was the government. The Club re-initiated a position in the Google parent in late December 2025 after a huge run higher. We built it up over time, and we're sitting on unrealized gains of 30%. We think that's only the beginning. Owning the past mistake and recognizing that Alphabet was on a winning path was not easy. But it was necessary. Back to Nvidia, from a valuation perspective, there really isn't much to debate. Compared to peers, the stock is clearly undervalued from a price-to-earnings standpoint. Forward earnings-based valuations, Broadcom (also a Club name) trades at 28 times, Marvell at 39 times, Advanced Micro Devices at 44 times, and Intel at 86 times. Nvidia clocks in at only 24 times forward earnings, despite being the name at the heart of the entire AI trade. It has more tentacles than any other company, thanks to a streak of high-profile investments into both key data center suppliers and customers. The issue, as we see it, is given the material valuation discount to peers, expectations for continued earnings growth, and an earnings catalyst on the near-term horizon, should you continue to wait before buying? Warren Buffett once said, "Investing is simple, but not easy." The simple part is the idea that you can teach pretty much anyone the mechanics of investing, or how to read financial statements or stock charts. What's not easy is controlling your emotions and maintaining discipline. That's arguably the single most important factor to being a successful investor over the long-term. I don't care who you are; it is not easy to watch a name you believe in lose over one-third of its value in a short period of time, as we saw with our cybersecurity holdings earlier this year. It's even more difficult to do what you know you must when the stock is doing the opposite of what you believe the fundamentals dictate, buy more. That's what we did. We bought CrowdStrike twice when the stock was being hit in February and March . The market has finally come around to our way of thinking that AI won't reduce the need for cybersecurity but increase it. On Monday, we trimmed CrowdStrike to protect some of those hard-fought gains. One way to keep your emotions in check is to examine the risk-reward. Nvidia is viewed by many as being a frontrunner to reach the fabled $10 trillion market cap, which implies some 82% upside from here. NVDA YTD mountain Nvidia YTD Technical analysis puts potential support somewhere in the $217 range (the old high), based on the Polarity Principle (old resistance once overcome, turns into support), which represents a little over 2% downside. So, now you need to consider your longer-term outlook. If you are with us in the camp that this name can be worth $10 trillion one day, then you aren't concerning yourself with trying to pick up the stock some 2% cheaper because you don't nitpick and try to be cute when it comes to a roughly 1:40 risk to reward ratio (which is what you're looking for when you expect to see support 2% below the current price but think shares could trade up over 80% in coming years), you can buy it. If the market gives you an even better setup a week from now (on the assumption we are making that Nvidia isn't about to derail the train they are driving), you buy that too because you've left yourself some room to do so. Even if the downside proves to be a bit more than expected, the relative valuation versus peers and versus Nvidia's own historical valuation range, along with the potential upside, dictates that you've got a solid margin of safety at current levels. With a small position, you can hope for rips to the upside and take solace in the fact that making some money is better than making nothing. On the other hand, if it moves lower, you can take solace in the idea that the near-term decline only serves to boost the longer-term gains thanks to the reduced overall cost basis you'll achieve by stepping in to buy the weakness. Bottom line The fundamental reasons to own Nvidia, and not trade it, are three-fold: It's cheaper than its peers, still growing rapidly, and is now breaking out from a long-period of consolidation. We have not heard a single comment all earnings season from hyperscalers to suggest that demand has done anything other than strengthen in recent months. And, with billions upon billions already invested, Alphabet , Amazon , Microsoft , and Meta Platforms revealed alongside earnings at least $695 billion of capital expenditure plans for this year, a 14% increase from the previous estimate of $608 billion. We expect Nvidia to get its fair share. The decision to buy Nvidia ahead of Wednesday evening's earnings, if you don't have a position, is up to you. But, considering our argument that Nvidia would be great in any portfolio, even at current prices, there is probably no wrong answer. (Jim Cramer's Charitable Trust is long NVDA, AVGO, GOOGL, AMZN, META, MSFT, CRWD. See here for a full list of the stocks.) As a subscriber to the CNBC Investing Club with Jim Cramer, you will receive a trade alert before Jim makes a trade. Jim waits 45 minutes after sending a trade alert before buying or selling a stock in his charitable trust's portfolio. If Jim has talked about a stock on CNBC TV, he waits 72 hours after issuing the trade alert before executing the trade. THE ABOVE INVESTING CLUB INFORMATION IS SUBJECT TO OUR TERMS AND CONDITIONS AND PRIVACY POLICY , TOGETHER WITH OUR DISCLAIMER . NO FIDUCIARY OBLIGATION OR DUTY EXISTS, OR IS CREATED, BY VIRTUE OF YOUR RECEIPT OF ANY INFORMATION PROVIDED IN CONNECTION WITH THE INVESTING CLUB. NO SPECIFIC OUTCOME OR PROFIT IS GUARANTEED.
Four leading AI models discuss this article
"Nvidia's valuation discount is less compelling once customer concentration and pre-earnings volatility are factored in."
The article frames NVDA as a buy at current levels due to 24x forward earnings versus higher multiples at AVGO, AMD, and others, plus AI-driven capex growth from hyperscalers. This glosses over the extreme concentration risk: Nvidia derives the bulk of its revenue from a handful of customers whose $695B spending plans could slow if AI returns disappoint. Buying into Wednesday's earnings also ignores how even strong beats have produced violent post-print swings when guidance fails to exceed sky-high expectations. The 2% technical support cited offers little cushion against a 10-15% drawdown if sentiment shifts.
Even if capex growth moderates, Nvidia's software and full-stack dominance could still drive outsized share gains, making the valuation discount to peers look conservative rather than misleading.
"Nvidia is cheaper than peers, but the valuation gap may exist because hyperscalers' in-house silicon efforts pose a structural threat the article dismisses."
The article conflates two separate questions: whether Nvidia is cheap (it is, at 24x forward P/E vs. peers at 28-86x) and whether that valuation gap persists for good reason. The $10 trillion market cap thesis requires not just AI adoption but Nvidia maintaining dominance despite AMD, Intel, and custom silicon efforts by hyperscalers. The article cites $695B capex plans as demand proof, but doesn't address that this spending increasingly funds *in-house* chip development (Amazon Trainium, Google TPU, Meta MTIA). The 2% downside support claim relies on technical analysis—the Polarity Principle—which is speculative. Most critically: the article was written pre-earnings, yet treats a beat-and-raise as 'minimum requirement,' implying the bar is already priced in. If NVDA meets expectations, the stock may not move; if it misses, the valuation discount evaporates fast.
Nvidia's 24x forward multiple may reflect justified caution: hyperscalers are actively building alternatives to reduce chip costs, and a slowdown in capex growth (from 14% YoY to single digits) would crater earnings. The article's 82% upside assumes perpetual market-share dominance that isn't guaranteed.
"Nvidia's valuation discount is a mirage that relies on unsustainable capex growth from hyperscalers who have yet to prove a concrete return on their AI investments."
Nvidia’s forward P/E of 24x looks deceptively attractive, but this valuation metric is highly sensitive to the 'denominator'—analyst earnings estimates that assume zero margin compression. The article ignores the risk of hyperscaler capex fatigue; while $695 billion in spending is bullish, it is also a massive hurdle to clear. If AI ROI doesn't materialize for Big Tech, that capex will be slashed, gutting Nvidia’s growth narrative. The '1:40 risk-reward' ratio is a classic trap; it assumes the $10 trillion market cap is a base case rather than a speculative ceiling. Investors should be wary of the 'cheap' valuation label when the entire sector is priced for perfection.
If Nvidia’s Blackwell and Rubin architectures maintain their current performance lead, the company effectively functions as a tax on the entire global compute infrastructure, rendering traditional valuation models obsolete.
"NVDA's upside depends on a continuing AI capex boom that may be episodic; a sustained slowdown in hyperscale data-center demand or regulatory/supply pressures could trigger a meaningful multiple contraction despite healthy earnings."
Article argues it's not too late to buy NVDA, citing a 'beat and raise' expectation, broad AI demand tailwinds; it also notes the stock has pulled back after a run. My take to stress-test: the bull case relies on a sustained AI capex cycle led by hyperscalers, but that cycle may be episodic and cyclical; Nvidia's valuation already prices in aggressive growth, leaving little margin for disappointment in Q2 guidance or-margin pressure from supply and R&D. Risks glossed over include potential demand slowdowns, competition from AMD/others in data-center GPUs, regulatory/export controls, and external macro shocks that hit IT budgets. The article also underplays the risk that an earnings miss could trigger a rapid multiple contraction.
The bullish setup rests on a perpetual AI cycle; a modest miss or softer guidance could trigger a sharp re-rating as spending slows and rivals steal share. If policy or supply chain constraints bite, Nvidia's moat could erode faster than expected.
"TSMC constraints delay custom silicon impact until 2026, protecting near-term Nvidia growth."
Claude flags hyperscaler in-house chips as a direct threat to Nvidia's dominance, but this underestimates TSMC capacity bottlenecks that will delay meaningful custom-silicon volume until at least 2026. Even if Amazon and Google ramp Trainium and TPU, training workloads still require Nvidia GPUs in the interim, keeping 2025 data-center growth above 50% and supporting the 24x multiple rather than forcing immediate compression.
"Custom silicon delays don't protect Nvidia's margins if hyperscalers shift capex allocation psychology today."
Grok's TSMC bottleneck argument assumes custom silicon delays don't matter until 2026, but misses the optionality effect: hyperscalers funding R&D today signals intent to reduce Nvidia dependency *eventually*. Even if volume lags, capex allocation shifts psychologically—CFOs begin budgeting for alternatives, pressuring Nvidia's pricing power in 2025-26 regardless of TSMC timelines. The 50% growth assumption doesn't account for this margin erosion.
"Hyperscaler efficiency gains and custom silicon development will lead to demand destruction for Nvidia GPUs sooner than the current consensus expects."
Claude and Grok are debating timing, but both ignore the 'inventory overhang' risk. If hyperscalers are building custom silicon, they are simultaneously optimizing existing workloads to require fewer cycles. This efficiency gain—often overlooked—is a direct headwind to Nvidia’s volume growth. Even if TSMC capacity keeps Nvidia relevant, the 'tax' on global compute Gemini mentioned is being actively engineered away by the very customers currently funding it. The valuation is not just about competition; it is about demand destruction.
"Regulatory headwinds could cap Nvidia's growth and compress multiples, undermining the in-house chip optionality thesis."
Claude's optionality argument ignores regulatory headwinds that could blunt demand for Nvidia GPUs in China and other restricted markets. Export controls and potential future sanctions limit addressable market, pressure pricing, and can force channel diversification into less profitable segments. Even with in-house accelerators, policy risk is a real ceiling on growth and could trigger multiple compression if earnings miss or guidance narrows. The moat may not obviate policy risk.
The panelists generally agreed that NVDA's valuation is attractive on the surface, but they raised significant concerns about the sustainability of its growth narrative and the risks associated with hyperscaler capex fatigue, in-house chip development, and regulatory headwinds. They also noted that NVDA's stock price may not move even with a beat-and-raise earnings report, and could face a rapid multiple contraction if guidance disappoints.
NVDA's 24x forward P/E multiple may support its stock price in the near term, given its dominance in data-center GPUs and the potential for strong earnings growth.
Hyperscaler capex fatigue and in-house chip development could gut NVDA's growth narrative and compress its valuation.