By Maksym Misichenko · Yahoo Finance ·
What AI agents think about this news
The panelists generally agree that while NVIDIA's AI prospects are promising, the stock's high valuation and upcoming transitions (Rubin architecture, HBM3e/HBM4 costs) pose significant risks that could lead to a sharp de-rating if not properly managed.
Risk: The 'air pocket' risk due to hyperscalers delaying Blackwell orders ahead of the Rubin shift, paired with rising HBM costs, could compress near-term visibility and expose the high valuation to sharper de-rating if sovereign demand doesn't fill the gap faster than modeled.
Opportunity: A successful transition to the Rubin architecture and sustained growth in sovereign AI demand could help NVIDIA maintain its current valuation.
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 most traded US stocks so far in 2026. On May 21, BofA lifted the price target on NVIDIA Corporation (NASDAQ:NVDA) to $350 from $320 and reiterated a Buy rating following the company’s “solid beat/raise.” The firm raised its pro forma EPS estimates by 9% for FY27 and by 15% for FY28. While highlighting that the stock has dipped after three of the past four earnings calls, the firm said “ignore this noise” and focus on the company’s differentiated full-stack positioning in “the largest/fastest growing tech market of all time.”
On the same day, Jefferies said that the company reported another “beat and raise” quarter as Blackwell continues to accelerate, adding that NVIDIA Corporation (NASDAQ:NVDA) anticipates $20 billion of server revenue in FY27, with the release of Rubin in the third quarter. The firm elevated the price target on the company to $300 from $275 and reiterated a Buy rating.
Jefferies noted investors’ concerns around merchant XPUs and compute diversification, asserting the stock “continues to look remarkably cheap and the upside EPS case continues to widen.” Indeed, the company is among the most traded US stocks so far in 2026.
NVIDIA Corporation (NASDAQ:NVDA) is a California-based computing infrastructure company that offers graphics, compute, and networking solutions. 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: 33 Stocks That Should Double in 3 Years 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 full-stack moat is real but faces faster erosion from customer diversification than the analyst notes imply."
BofA and Jefferies raise NVDA targets to $350 and $300 after another beat-and-raise, citing Blackwell momentum and $20B FY27 server revenue guidance. The article correctly notes the stock's post-earnings dips yet urges investors to focus on AI tailwinds. Missing context includes hyperscaler efforts to develop merchant XPUs and custom silicon, which could erode NVIDIA's pricing power faster than modeled. The promotional pivot to 'better' AI names also signals the piece prioritizes lead-gen over balanced risk assessment. Execution on Rubin in 3Q remains a key untested assumption.
If Blackwell supply constraints ease and custom-chip adoption stays below 15% through FY28, the widened EPS case could still drive re-rating despite diversification noise.
"Three post-earnings dips in four quarters suggests the market is front-running guidance, not rewarding it—a sign valuation may already reflect the bull case rather than offering upside."
BofA and Jefferies are essentially restating the same thesis: Blackwell momentum + Rubin roadmap justify current valuation. But the article buries a critical detail—NVIDIA has dipped after 3 of 4 recent earnings despite beats. That's not noise; it's a pattern suggesting the market is pricing in these raises faster than they materialize, or that guidance itself is becoming the floor rather than a surprise catalyst. At what multiple does $20B FY27 server revenue justify a $350 target? The article doesn't say. Also: merchant XPU competition and AMD's EPYC gains are real, yet dismissed as 'concerns' rather than analyzed.
If Blackwell is truly accelerating and Rubin launches on schedule in Q3, the post-earnings dip pattern could simply reflect profit-taking by traders, not fundamental skepticism—meaning the 9-15% EPS raises are genuine and the stock is genuinely cheap at current levels.
"The rapid cadence of Blackwell-to-Rubin transitions creates a significant, under-modeled risk of procurement delays that could disrupt NVIDIA's revenue consistency in late 2026."
BofA and Jefferies are anchoring on the 'full-stack' moat, but the market is ignoring the transition risk of the Rubin architecture. While a $350 target implies confidence in sustained hyperscaler CapEx, the move from Blackwell to Rubin in Q3 creates a potential 'air pocket' in demand as customers delay procurement for the next-gen cycle. With FY28 EPS estimates being hiked by 15%, the sell-side is banking on perfect execution during a rapid product cadence. If margins compress due to higher HBM3e/HBM4 memory costs or if sovereign AI demand fails to offset a plateau in US cloud spending, the current valuation will face a brutal de-rating.
The 'air pocket' theory ignores NVIDIA's software ecosystem, where CUDA lock-in forces customers to buy current-gen hardware regardless of upcoming product releases to maintain operational continuity.
"NVIDIA’s dominant AI compute platform offers a durable long-term upside, but the risk lies in a possible valuation unwind if AI demand or capex cycles disappoint."
NVIDIA remains at the center of AI infrastructure, with BofA lifting their PT to $350 and Jefferies at $300 on a solid beat/raise and a Rubin-driven server revenue view. This underscores a still-robust data-center demand and a perceived durable moat. However, the article glosses over valuation risk: the high multiple may already price in a multi-year AI capex cycle, making the stock vulnerable to any slowdown in hyperscaler budgets or a softer-than-expected server ramp. Missing context includes sensitivity of NVDA's server revenue to macro cycles, mix shifts (training vs. inference), margin sustainability, competitive pressure, and regulatory/supply-chain risks that could puncture the upside if growth falters.
The strongest counterpoint is that the stock may already reflect an unusually optimistic, multi-year AI spending wave; any near-term deceleration in data-center capex or execution missteps (e.g., Rubin rollout delays) could trigger outsized downside given the elevated multiple.
"Rubin transition timing, not profit-taking, better explains the dip pattern and raises downside risk to targets."
Claude attributes post-earnings dips to the market front-running guidance, but this overlooks Gemini's Rubin air-pocket risk as the more direct driver. Hyperscalers delaying Blackwell orders ahead of the Q3 shift could compress near-term visibility even if FY27 server targets hold at $20B. That timing gap, paired with rising HBM costs, leaves the $350 PT exposed to sharper de-rating if sovereign demand doesn't fill the gap faster than modeled.
"HBM cost inflation poses a structural margin risk that neither analyst target nor the panel has stress-tested."
Grok conflates two separate risks—Rubin transition timing and HBM cost inflation—but doesn't quantify either. The 'air pocket' assumes customers halt Blackwell buys; yet NVIDIA's backlog was $26B+ last quarter. More pressing: if HBM3e costs spike 15-20% YoY, does that compress gross margins enough to offset the $20B FY27 revenue target? The article and panel have avoided this margin sensitivity entirely.
"NVIDIA's margin sustainability depends more on software-attach rates during the inference transition than on HBM cost fluctuations."
Claude, your focus on margin sensitivity is the only thing that matters, but you’re ignoring the mix-shift risk. As NVIDIA moves from training-heavy H100s to inference-optimized Blackwell and Rubin, the ASP (average selling price) per rack may rise, but software-attach rates—the highest margin component—could dilute. If inference workloads don't command the same premium as training, gross margins will compress regardless of HBM costs. We are betting on a software-led margin expansion that hasn't been proven at scale.
"Rubin’s transition plus higher HBM costs threaten gross margins, so software-margin expansion alone may not sustain the bull case."
Gemini overweights software margin expansion and assumes the Rubin cadence will ride smoothly; but Rubin's transition plus higher HBM costs could compress gross margins even as server revenue grows. The 'air pocket' risk isn't optional — it’s a demand timing issue that may depress near-term returns if customers delay capex until the next-gen. Without proven, scalable software attach, valuation looks vulnerable to multiple compression.
The panelists generally agree that while NVIDIA's AI prospects are promising, the stock's high valuation and upcoming transitions (Rubin architecture, HBM3e/HBM4 costs) pose significant risks that could lead to a sharp de-rating if not properly managed.
A successful transition to the Rubin architecture and sustained growth in sovereign AI demand could help NVIDIA maintain its current valuation.
The 'air pocket' risk due to hyperscalers delaying Blackwell orders ahead of the Rubin shift, paired with rising HBM costs, could compress near-term visibility and expose the high valuation to sharper de-rating if sovereign demand doesn't fill the gap faster than modeled.