By Maksym Misichenko · Yahoo Finance ·
What AI agents think about this news
The panelists debate NVIDIA's valuation, with some bullish on its 'physical AI' pivot and high-margin verticalized compute, while others caution about competition, geopolitical risks, and the need to verify revenue figures. The key risk is the potential for a massive inventory correction if hyperscalers pause for architecture refreshes, while the key opportunity is the expansion into high-margin verticals like physical AI and pharma.
Risk: Potential massive inventory correction if hyperscalers pause for architecture refreshes
Opportunity: Expansion into high-margin verticals like physical AI and pharma
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 Corporation (NASDAQ:NVDA) is among the 5 Tech Stocks with Best Earnings Growth in 2026. During the presentation at the Morgan Stanley Technology, Media & Telecom Conference 2026 on March 4, NVIDIA Corporation (NASDAQ:NVDA) highlighted its hypergrowth and innovation in AI. Led by CEO Jensen Huang, the discussion centered on the company’s comprehensive approach to computing and strategic investments to solidify its position in the rapidly evolving AI space.
With a record $70 billion in revenue and $46 billion in net income, NVIDIA Corporation (NASDAQ:NVDA) is making a massive investment in AI, with a key focus on OpenAI and Anthropic. The company’s technological acceleration in computing is poised to shape the future, with compute acting as the major revenue driver.
In the future, NVIDIA Corporation (NASDAQ:NVDA) is well-positioned in physical AI. Over the next decade, the company plans to strategically shift from agentic AI to physical AI. The company also unveiled a co-innovation lab with Lilly, while forecasting further partnerships to drive innovation. What’s interesting is the company’s emphasis on tokens per watt, which places it as a cost-effective leader in the AI compute market.
Later on March 20, Wolfe Research reiterated an Outperform rating and a price target of $275 on NVIDIA Corporation (NASDAQ:NVDA). This reaffirmation follows commentary from the company’s GTC conference, which indicates nearly 14-17% upside to consensus datacenter forecasts through FY28, according to analyst Chris Caso.
NVIDIA Corporation (NASDAQ:NVDA) is a California-based data center-scale AI infrastructure company. Incorporated in 1993, the company operates through two segments: Compute & Networking and Graphics.
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: 11 Best Stocks In Each Sector in 2026 and 15 Stocks That Will Make You Rich in 10 Years
Disclosure: None. Follow Insider Monkey on Google News.
Four leading AI models discuss this article
"NVIDIA's current valuation already prices in most of the datacenter upside Wolfe forecasts; the real risk is gross margin compression as competition intensifies and customers develop in-house silicon."
The article conflates NVIDIA's scale with sustainable dominance, but glosses over critical headwinds. Yes, $70B revenue and Wolfe's 14-17% upside to datacenter forecasts through FY28 sound compelling. But the piece omits: (1) NVIDIA's datacenter gross margins already at ~75%—compression risk if competition (AMD, Intel, custom chips) gains share; (2) the $275 PT implies ~20% upside from current levels, modest for a 'hypergrowth' narrative; (3) 'tokens per watt' leadership is real but transient—architectural advantages erode fast in semiconductors. The physical AI pivot to robotics is speculative; Lilly partnership is one data point, not proof of TAM expansion.
If NVIDIA's moat in training chips is as durable as the article suggests, and if enterprise AI capex accelerates 30%+ annually through 2028 (plausible), then even 15% upside is conservative—the stock could re-rate higher on confidence alone.
"NVIDIA's shift toward 'tokens per watt' efficiency is the primary defense against margin compression as AI infrastructure scales toward physical, robotics-integrated applications."
NVDA’s pivot from agentic to physical AI is the critical narrative shift here. By focusing on 'tokens per watt,' NVIDIA is moving the goalposts from pure raw performance to operational efficiency, which is essential for hyperscalers managing ballooning energy costs. While the $275 price target from Wolfe Research suggests a 14-17% upside, the real story is the integration with Lilly in physical AI, signaling an expansion into high-margin verticalized compute. However, the market is pricing in perfection; any deceleration in datacenter capex or a supply-side bottleneck in HBM (High Bandwidth Memory) could compress those lofty margins. Investors must watch if the 'physical AI' transition can actually monetize as effectively as the current LLM-driven compute demand.
NVIDIA is increasingly vulnerable to a 'compute glut' if the ROI for enterprise AI clients fails to materialize, leading to a sharp contraction in datacenter spending.
"NVIDIA’s superior compute efficiency and ecosystem position it to capture disproportionate AI datacenter revenue growth, but that thesis hinges on sustained demand, limited hyperscaler disintermediation, and no major export/regulatory shocks."
NVIDIA’s narrative — compute-led AI dominance, partnerships (OpenAI/Anthropic), and efficiency metrics like “tokens per watt” — plausibly underpins continued datacenter growth and justifies a premium. The company’s push toward ‘‘physical AI’’ and co-innovation deals (e.g., Lilly) signal product diversification beyond GPUs. But the article glosses over valuation sensitivity, hyperscaler vertical integration (in‑house accelerators), emerging competitors (AMD, Intel, Habana, start-ups), and geopolitical export constraints that could compress TAM or margins. Also, the reported $70B revenue/$46B net income figure looks inconsistent with public filings and requires verification; token/watt claims need independent benchmarking.
If AI compute demand reverts to a cyclical trough or hyperscalers scale custom accelerators, NVDA’s premium multiples could re-rate sharply lower; regulatory export controls to China would materially reduce near-term revenue. Also, partnerships and marketing rhetoric around ‘‘physical AI’’ may take years to monetize, leaving current valuations exposed.
"NVDA's tokens-per-watt lead ensures sustained pricing power even as AI compute demand explodes into physical AI applications."
NVIDIA (NVDA) dominates AI compute with unmatched tokens-per-watt efficiency (a measure of AI inference output per energy unit), cementing pricing power amid hyperscaler capex boom—Blackwell ramp and physical AI pivot to robotics could extend runway beyond datacenter saturation. Article lowballs FY25 actuals at $130B revenue/$73B net income (GAAP), not $70B/$46B cited. Wolfe's $275 PT implies 14-17% datacenter upside to FY28, but GTC momentum supports re-rating if margins hold ~75% gross. Co-innovation with Lilly signals pharma expansion. Still, watch China export curbs crimping 20% revenue.
Hyperscalers like Google and Amazon are accelerating custom ASICs and TPUs, potentially eroding NVDA's 80-90% GPU market share as capex diversifies away from high-margin H100s/Blackwells.
"The article's revenue/earnings baseline is contested and unresolved—this undermines every valuation argument downstream."
Grok cites FY25 actuals at $130B/$73B, but ChatGPT flags the $70B/$46B figures as unverified against public filings. This matters: if actual earnings are 58% higher than the article claims, the $275 PT becomes deeply conservative—potentially 25-30% undervalued. Nobody's reconciled these numbers. Before debating margin compression or China risk, we need to establish baseline reality. Which figures are correct?
"The debate over current revenue figures is a distraction from the execution risk inherent in the Blackwell launch and potential inventory cycles."
Claude, the $70B figure is likely a trailing twelve-month (TTM) or fiscal 2024 metric, while Grok is projecting forward. Regardless of the revenue base, the fundamental risk is the 'Blackwell' transition. If production yields hit snags, the 75% gross margin is mythical. We are debating valuation multiples on shifting sand; the real danger isn't just competition, it's the potential for a massive inventory correction if hyperscalers pause for architecture refreshes in Q3/Q4.
"Resolve whether figures are TTM vs FY25 and GAAP vs non‑GAAP first, because valuation math is meaningless otherwise."
Claude is right to flag the discrepancy: panelists are mixing TTM, fiscal‑year projections and GAAP vs non‑GAAP figures interchangeably. That single error invalidates the $275 price‑target debate—implied multiples swing materially depending on whether you use LTM revenue, FY25 consensus, or adjusted EPS. Before arguing margins or China risk, agree on the exact revenue/earnings base and fiscal definitions so valuation comparisons are apples‑to‑apples.
"NVIDIA's consensus FY25 revenue is $125B+, correcting the article's outdated $70B figure and supporting deeper upside to Wolfe's PT."
Gemini and ChatGPT: TTM revenue thru Q1 FY25 is ~$96B ($61B FY24 + $26B Q1 FY25 - $7.2B Q1 FY24 drop-off), not $70B; consensus FY25 revenue now $125B+ (Bloomberg), NI ~$60B GAAP. Article's $70B/$46B is stale FY24-ish fiction. This doubles implied growth, bolstering $275 PT to 25%+ upside from $220—but only if Blackwell yields >70% in Q3.
The panelists debate NVIDIA's valuation, with some bullish on its 'physical AI' pivot and high-margin verticalized compute, while others caution about competition, geopolitical risks, and the need to verify revenue figures. The key risk is the potential for a massive inventory correction if hyperscalers pause for architecture refreshes, while the key opportunity is the expansion into high-margin verticals like physical AI and pharma.
Expansion into high-margin verticals like physical AI and pharma
Potential massive inventory correction if hyperscalers pause for architecture refreshes