By Maksym Misichenko · Nasdaq ·
What AI agents think about this news
The panel consensus is that NVDA's $500B backlog for 2025-26 is not a guarantee of future revenue due to the contingent nature of hyperscaler capex and the potential for demand substitution towards custom ASICs. The panelists also highlighted the risks associated with circular financing and the cyclical nature of the semiconductor industry.
Risk: Demand substitution towards custom ASICs for inference, which could lead to margin compression and delay in backlog conversion.
Opportunity: The potential for strong AI hardware demand, as indicated by the $500B backlog.
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 →
Investors have long been bullish on Nvidia (NASDAQ: NVDA) for many reasons, but one that may not get a lot of direct attention is its backlog. Amid the massive demand for its AI accelerators, it accumulated a backlog of $500 billion in combined bookings for 2025 and 2026. Also, when CEO Jensen Huang was in Seoul in June, he stated that we are at the beginning of the AI boom, even as Nvidia stock has made massive gains over the last four years.
However, semiconductor stocks, including the chip giant, have pulled back in recent trading sessions as concerns about high capital expenditures (capex) and circular financing have given investors pause. Despite those concerns, investors should consider using this pullback to add shares. Here's why.
Missed Nvidia in 2009? This Rare Signal Is Flashing Again. In 2009, a "Double Down" signal flashed for a little-known chipmaker called Nvidia. For the first time in years, that same "Total Conviction" signal is flashing for a company 1/100th the size of Nvidia. Continue »
Admittedly, the uncertainty surrounding Nvidia is understandable. The estimated $725 billion in planned capex by hyperscalers could raise doubts about access to credit. Moreover, Nvidia is an investor in many of its customers, leading to questions about whether the $500 billion is a true reflection of the backlog.
Nonetheless, other factors cast doubt as to whether Nvidia is really in trouble. Huang added that investors should be "very happy" since they can "buy at a discount." He appears to have a point. Its stock sells at a P/E ratio of 30, barely above the S&P 500 average of 29.
Despite that valuation and Huang's assertion about the AI boom, Nvidia is not trading at a premium. In the first quarter of fiscal 2027 (ended April 26), its revenue of $81.6 billion rose by 85% from year-ago levels. That led to $58.3 billion in net income, a 211% annual increase. Also, since revenue surged 65% higher in fiscal 2026, its quarterly growth is more than a one-time event.
Considering Nvidia's growth rate, it likely does not need to artificially inflate its growth. Additionally, the "circular financing" is a reflection of its AI needs.
For example, it is an investor in the neocloud company CoreWeave, which is also a customer. In a technical sense, Nvidia does not "need" CoreWeave to grow.
Still, CoreWeave gives Nvidia a platform where it can more quickly deploy and test its hardware. Also, since the neocloud industry is new, the leading companies do not have the backing to scale quickly on their own.
With Nvidia's support, they can build faster, facilitating faster adoption of its technology. This means that such investments could pay off for Nvidia longer-term and, ultimately, ease the concerns of investors.
Despite doubts, Nvidia's backlog is a reason to buy its stock.
Indeed, concerns about credit accessibility could slow growth and discourage some investors from buying its stock. Also, its so-called "circular financing" makes it look like Nvidia has to buy its growth.
However, the stock's P/E ratio seems to confirm that it sells at a discount. Also, it appears investments like its position in CoreWeave contribute to the improvement and rapid deployment of its hardware.
Thus, despite the concerns, investors should continue to believe in Nvidia's growth. Considering its comparatively low P/E ratio, investors have good reason to look past the uncertainty and take a position in the chip stock.
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 $386,727! Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,232,139!
Now, it’s worth noting Stock Advisor’s total average return is 906% — 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 August 2, 2026. *
Will Healy has positions in CoreWeave. The Motley Fool has positions in and recommends 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
"The backlog is genuine visibility but its convertibility hinges on hyperscaler capex discipline that the article underweights."
The $500B backlog for 2025-26 is real and provides rare multi-year visibility in semiconductors, supporting continued 60-80% revenue growth if hyperscaler capex holds. NVDA's 30x P/E on 85% revenue and 211% NI growth does look reasonable versus historical AI leaders, and Jensen's 'beginning of the AI boom' comment aligns with still-rising inference and training demand. However, the article downplays concentration risk: four hyperscalers drive ~45% of revenue, and any slowdown in their $725B planned capex (already showing signs of scrutiny in 2H25 budgets) could crater the backlog conversion. Circular financing via stakes in CoreWeave et al. blurs true end-demand and creates correlated credit risk the piece dismisses too lightly.
If hyperscalers face higher-for-longer rates and ROI pressure on AI projects, the $500B backlog could see material cancellations or deferrals beyond 2026, exposing that NVDA trades at 45x cash earnings once capex-fueled growth normalizes.
"The $500 billion backlog is a non-binding projection that masks the severe risk of a sudden, synchronized capex contraction among major hyperscalers."
The article’s reliance on a $500 billion backlog is a dangerous metric for retail investors. While Nvidia’s revenue growth is undeniable, a backlog in the semiconductor space is not a binding contract; it is a non-binding reservation that can be canceled if hyperscaler demand shifts or if internal ROI metrics for AI infrastructure turn negative. The P/E ratio comparison to the S&P 500 is intellectually dishonest because it ignores the cyclical nature of hardware and the massive capex burden Nvidia’s customers are carrying. If Microsoft, Meta, and Alphabet decide to throttle spending in 2026, that $500 billion 'cushion' evaporates instantly, leaving Nvidia exposed to massive inventory write-downs.
If the AI transition is truly a foundational shift comparable to the internet, then current capex is not 'spending' but 'capital investment' that will yield multi-year productivity gains, justifying the current valuation.
"The $500B backlog is optionality, not revenue—and Nvidia's 30x P/E is premium, not discounted, when adjusted for cyclical risk and the working-capital cost of funding customer growth."
The $500B backlog is real optionality, but the article conflates two separate problems. First: the backlog's quality matters enormously. If $300B of that is contingent on hyperscalers securing financing or hitting ROI thresholds they haven't yet proven, it's not revenue—it's intent-to-buy. Second: the P/E of 30 is NOT a discount when normalized for semiconductor cyclicality. Nvidia trades at 30x on 85% YoY growth; that's actually premium to historical AI-cycle comps. The circular financing concern is dismissed too casually—if Nvidia must fund its own customers' capex to realize bookings, that's a working capital drain and implicit leverage that traditional P/E doesn't capture.
If the backlog converts at even 70% and AI capex actually accelerates (not decelerates), the stock could re-rate to 40x earnings within 18 months, making current valuation look cheap in hindsight.
"Backlog alone drives valuation only if conversion and cash flow stay robust; otherwise macro or financing shocks could re-rate Nvidia."
NVDA's $500B backlog for 2025–26 underscores strong AI hardware demand, but bookings are not cash and may not translate to immediate revenue. The article frames circular financing and hyperscaler capex as supportive, yet a downturn in credit conditions or a pullback in AI spending could push bookings out or reduce pricing power. CoreWeave and other financing links could lift deployment speed, but they also concentrate counterparty risk and imply a financing channel that might tighten in a cyclical downturn. Even with a lofty growth trajectory and a ~30x earnings multiple, the real test is backlog-to-revenue conversion and free cash flow stability if the AI cycle cools.
Backlog can be revised downward; circular financing may mask credit risk and push revenue recognition later than investors assume. A cyclical slowdown could compress margins if AI spending stalls or financing costs rise.
"Backlog conversion risk is amplified by correlated credit and supplier inventory dynamics that no panelist has quantified."
Claude's 70% backlog conversion scenario still understates second-order risk: if hyperscalers' $725B capex faces even modest 15-20% cuts in 2H25 due to ROI scrutiny, the resulting order cancellations could cascade into supplier inventory builds across the entire AI stack, not just NVDA. Circular financing doesn't merely add leverage; it synchronizes credit events across counterparties in a way traditional cycle models ignore.
"The primary risk is a supply-side glut caused by over-capacity commitments and a shift toward custom ASIC alternatives, rather than just hyperscaler demand volatility."
Grok and Claude are fixated on the demand side, but you are all ignoring the supply-side bottleneck: TSMC’s CoWoS capacity. Even if hyperscalers want to cut capex, they can't easily cancel orders because they are locked into long-term capacity commitments. The real risk isn't just 'ROI scrutiny'—it's the potential for a massive supply glut if NVDA over-orders capacity while end-user demand shifts toward custom silicon (ASICs) from Broadcom or Marvell, which offer better TCO for inference.
"Supply-side rigidity at TSMC locks in NVDA's capex exposure, but accelerates the shift toward inference-optimized custom silicon that erodes blended ASP."
Gemini's CoWoS constraint is the critical miss in our demand-side debate. TSMC's multi-year capacity commitments mean NVDA can't easily right-size orders even if hyperscaler ROI craters. But this cuts both ways: if custom silicon (Broadcom ASICs) gains traction for inference, NVDA's backlog converts to lower-margin training chips only. That's margin compression, not revenue collapse—but it's the real tail risk nobody quantified.
"The real risk to NVDA’s upside is demand substitution to ASICs for inference, which could erode backlog quality and compress margins despite supply constraints."
Responding to Gemini: supply tightness via CoWoS is real, but the bigger, underappreciated risk is demand substitution toward custom ASICs for inference. If Broadcom/Marvell-style ASICs or end-to-end inference stacks gain traction, hyperscalers may convert or degrade NVDA’s backlog into lower-margin gear, or delay ramp even with capacity constraints. In that world, CoWoS bottlenecks actually help NVDA price discipline in the near term, but they don’t immunize the stock from multi-year margin compression if ASICs scale.
The panel consensus is that NVDA's $500B backlog for 2025-26 is not a guarantee of future revenue due to the contingent nature of hyperscaler capex and the potential for demand substitution towards custom ASICs. The panelists also highlighted the risks associated with circular financing and the cyclical nature of the semiconductor industry.
The potential for strong AI hardware demand, as indicated by the $500B backlog.
Demand substitution towards custom ASICs for inference, which could lead to margin compression and delay in backlog conversion.