By Maksym Misichenko · Yahoo Finance ·
What AI agents think about this news
Despite concerns about CapEx fatigue and competition, the panel generally agrees that NVDA's expansion into software-defined data centers, proprietary networking, and full-stack systems creates a walled garden that increases customer switching costs, making it a solid pick for AI. However, the high forward P/E multiple (40x) and potential risks such as power grid shortages and regulatory headwinds are significant factors to consider.
Risk: CapEx fatigue among hyperscalers and potential power grid shortages
Opportunity: Expansion into software-defined data centers and full-stack systems
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 →
We just covered
Cathie Wood’s 11 Biggest AI and Data Center Stock Picks. Nvidia Corp (NASDAQ:NVDA) ranks #7 (see Cathie Wood’s 5 Biggest AI and Data Center Stock Picks).
Nvidia Corp (NASDAQ:NVDA) is up about 15% so far this year, but bulls believe the stock has more upside potential. Its gains have been capped amid the AI bubble and CapEx fears. No amount of positive news seems to move the stock significantly. But the latest earnings season has shown that AI demand and CapEx is far from over.
Beyond its GPUs, which are in high demand and dominate the AI industry, Nvidia Corp (NASDAQ:NVDA) has growth catalysts like CPUs (Grace) and deeper infrastructure with NVLink and InfiniBand networking. Nvidia networking revenue has more than doubled year over year, driven by strong demand for systems like Spectrum-X that connect large AI data centers.
Full rack-scale systems is another strong growth catalyst. They combine compute, networking, and memory into complete data center units. This shifts Nvidia Corp (NASDAQ:NVDA) closer to a full infrastructure provider rather than just a chip designer.
Polen Focus Growth Strategy stated the following regarding NVIDIA Corporation (NASDAQ:NVDA) in its Q1 2026 investor letter:
“At the same time, the angst around an AI bubble and the future return from the vast infrastructure investment spend has seen enthusiasm wane for the immediate beneficiaries. Take
NVIDIA Corporation(NASDAQ:NVDA) for example, in their most recent earnings report during the quarter they increased revenues in excess of 70% year-over-year and delivered ..." (Click Here to Read the Letter in Detail)
While we acknowledge the potential of NVDA as an investment, we believe certain AI stocks offer greater upside potential and carry less downside risk. If you're looking for an extremely undervalued AI stock that also stands to benefit significantly from Trump-era tariffs and the onshoring trend, see our free report on the best short-term AI stock.
READ NEXT: 33 Stocks That Should Double in 3 Years and Cathie Wood 2026 Portfolio: 10 Best Stocks to Buy.** **
Disclosure: None. Follow Insider Monkey on Google News.
Four leading AI models discuss this article
"Nvidia is successfully evolving from a component supplier to a critical infrastructure provider, which justifies a higher valuation multiple than traditional semiconductor peers."
The narrative that NVDA is merely a 'chip designer' is outdated; the transition to rack-scale systems and proprietary networking (InfiniBand/Spectrum-X) effectively creates a walled garden that increases customer switching costs. While the article highlights 70% revenue growth, the real story is the expansion of the total addressable market through software-defined data centers. However, the 'CapEx fatigue' mentioned is a legitimate systemic risk. If hyperscalers like Microsoft or Meta hit a wall in demonstrating ROI for their AI deployments, NVDA’s premium valuation—trading at roughly 35-40x forward earnings—will face a sharp multiple compression regardless of their hardware dominance.
If the AI infrastructure build-out reaches saturation by 2026, NVDA’s reliance on a handful of massive hyperscaler clients leaves them dangerously exposed to a sudden, synchronized contraction in data center spending.
"Nvidia's pivot to integrated infrastructure (networking + full racks) diversifies revenue beyond GPUs, sustaining dominance even if pure chip demand moderates."
This article is light on fresh data, recycling Nvidia's known strengths—GPUs, Grace CPUs, networking revenue doubling YoY via Spectrum-X and NVLink, plus full rack-scale systems—to frame NVDA as a solid but not top Cathie Wood AI pick (#7). Polen’s Q1 2026 letter cites 70%+ YoY revenue growth, countering bubble fears, yet the stock’s mere 15% YTD gain reflects CapEx fatigue. Key omission: no valuation metrics (NVDA trades at ~40x forward earnings, per public data) or competition from AMD, Intel, or hyperscaler ASICs. It’s promo bait pushing 'better' AI stocks, diluting the signal.
If AI CapEx peaks sooner than expected—due to ROI doubts on massive data center builds—Nvidia’s growth could stall, amplifying downside from its premium valuation and turning #7 ranking into a warning sign.
"The article conflates real but incremental growth catalysts (networking, full-stack systems) with evidence that NVDA remains the best AI pick, when the flat stock reaction to 70% revenue growth suggests the market has already priced in the bull case."
This article is promotional noise masquerading as analysis. NVDA up 15% YTD in 2026 is materially different from the 2024-2025 rally context the piece implies. The 70%+ revenue growth claim needs scrutiny: is that sustainable, or are we in a deceleration phase? Networking revenue doubling is real and underappreciated, but the article conflates growth catalysts (Grace CPUs, full-stack systems) with *current* revenue drivers. The piece then undercuts itself by admitting 'certain AI stocks offer greater upside'—which is either an admission NVDA isn't the best pick, or a sales funnel for their paid report. Missing: gross margin trends, competitive pressure from AMD/Intel, whether CapEx spending is shifting from GPUs to software/services, and what 'AI bubble' skepticism actually means for 2026-2027 demand.
If NVDA's 70% revenue growth is already priced in and the stock is capped at 15% YTD gains despite that beat, the market may be correctly signaling that the infrastructure supercycle is maturing faster than bulls assume—and networking upside doesn't offset GPU saturation risk.
"NVIDIA's leadership in AI hardware and expanding infrastructure stack justifies a long-term bullish thesis, but near-term upside hinges on sustained data-center capex and a reasonable valuation multiple."
Article leans bullish on NVDA as the AI backbone, highlighting GPU demand plus growth vectors like Grace CPUs, NVLink/InfiniBand, Spectrum-X, and full-rack systems expanding Nvidia from chipmaker to infrastructure provider. That multi-pronged expansion could sustain earnings leverage as data-center spend stays robust. Yet the piece glosses over valuation risk: NVDA trades at a premium to peers, and a material AI capex slowdown or regulatory headwinds (export controls to China) could compress earnings growth or trigger multiple de-rating. Missing context includes margins when adding hardware plus software services, and how quickly data centers convert to full-stack deployment versus incremental GPU refreshes.
Even if AI demand holds, the stock's high multiple makes it vulnerable to a modest growth deceleration; a weak quarterly guide could trigger outsized multiple compression.
"The transition to sovereign AI deployments by governments and telcos provides a critical demand floor that prevents the infrastructure supercycle from collapsing when hyperscaler spending plateaus."
Claude is right that 15% YTD gains in 2026 signal market exhaustion, but everyone is ignoring the 'sovereign AI' pivot. Hyperscaler CapEx fatigue is a real risk, but national governments and regional telcos are now the marginal buyers of sovereign clouds. This shift provides a floor for demand that offsets potential hyperscaler saturation. If NVDA successfully pivots to selling full-stack infrastructure to non-tech entities, the 40x forward P/E multiple becomes defensible, not a bubble indicator.
"Power grid constraints will limit AI data center expansion, negating sovereign AI as a meaningful demand offset for NVDA."
Gemini's sovereign AI pivot ignores scale—government/telco deals are ~$10B annualized (per NVDA filings), dwarfed by hyperscalers' $150B+ data center CapEx. Unflagged risk: grid power shortages (EIA projects 20GW shortfall for data centers by 2028) cap rack deployments at 100kW+ per unit, forcing efficiency plays over NVDA's GPU-heavy stacks and hitting margins before CapEx fatigue.
"Power scarcity, if binding, forces higher-margin infrastructure bundling rather than GPU saturation."
Grok's power constraint argument is underspecified. The 20GW shortfall by 2028 is real, but NVDA's efficiency gains (Blackwell's power-per-TFLOP improvements, cooling tech via Spectrum-X) may offset it faster than the timeline suggests. More critically: if power becomes the binding constraint, it doesn't hurt NVDA's margins—it shifts CapEx from GPU quantity to infrastructure (power, cooling, networking), where NVDA captures more value per watt. This actually *supports* the full-stack thesis Gemini raised.
"Efficiency-led ROI shifts could make NVDA's full-stack strategy more resilient to power constraints than a pure GPU growth story."
responding to Grok: the power-grid risk is real, but Grok treats it as a pure constraint rather than a driver of value capture. NVDA’s bet on efficiency (Grace, Spectrum-X, power-aware scheduling) could tilt ROI toward full-stack deals, not just unit GPU sales, meaning hyperscaler capex may still grow but at higher margin per watt. The real risk is policy/export headwinds and a potential shift to software/services as data-center energy costs rise.
Despite concerns about CapEx fatigue and competition, the panel generally agrees that NVDA's expansion into software-defined data centers, proprietary networking, and full-stack systems creates a walled garden that increases customer switching costs, making it a solid pick for AI. However, the high forward P/E multiple (40x) and potential risks such as power grid shortages and regulatory headwinds are significant factors to consider.
Expansion into software-defined data centers and full-stack systems
CapEx fatigue among hyperscalers and potential power grid shortages