Rosenblatt raises Nvidia stock price target on AI demand growth

Rosenblatt lifted its Nvidia (NVDA) price target to $390 (Buy) as strong AI demand supported bumper results. Management guided fiscal 2028 revenue growth to 70% YoY (above ~45% consensus) and said its AI platform segment should outpace hyperscaler growth longer term, while multi-year purchase commitments rose ~135% QoQ to $279B (mainly memory). Nvidia’s Vera Rubin architecture is projected to drive 20% of 3Q FY2027 data center revenue, and the company returned $26B to shareholders via $20B buybacks and $6B dividends (60% of FCF); shares rise on the outlook and momentum.

Analysis

This is less a one-quarter beat than evidence that AI infrastructure is moving from a hyperscaler-only story to a broader capex cycle. The important second-order effect is that NVDA is increasingly the toll collector on the whole stack: as demand visibility extends, the bottleneck migrates toward memory, optics, and power delivery, which should favor names like COHR and the memory complex while squeezing weaker AI cloud entrants that have to fund inventory up front.

Near term, the stock can keep drifting higher because estimate revisions will lag the underlying order signal, but the cleaner upside may actually sit in suppliers with lower expectations. Over 1-3 months, the key catalyst is whether hyperscalers and sovereign/enterprise buyers validate the demand curve in their own capex commentary; over 6-18 months, compute-per-watt gains could extend GPU replacement cycles and keep NVDA structurally share-gaining even if unit growth normalizes.

The contrarian risk is that the market is extrapolating demand strength faster than the supply chain can clear it. A memory-constrained ramp can keep headline revenue strong while delaying deployments, which is how momentum names get hit: not on a bad quarter, but on a slower-than-expected conversion from backlog to shipments. Falsifiers are a guide that misses by low-single digits, any gross margin compression from mix or competition, or evidence that the next architecture ramp is slower than promised.

Key Decisions for Investors

• Buy NVDA on 3%-5% post-rally pullbacks, or use a 3-6 month call spread to reduce implied-vol risk; thesis is continued analyst revision upside and durable demand visibility, with the trade invalidated if forward guidance fails to beat consensus or margins compress.
• Go long COHR as the cleaner second-order beneficiary of AI interconnect and optical intensity; hold for 1-3 months into next capex commentary cycle, and cut if AI networking spend shows signs of normalization.
• Pair long NVDA / short AMD for 1-2 earnings cycles to express share-gain and ecosystem lock-in; this is a relative-value trade, not a market call, and should be exited if AMD shows an inflection in accelerator traction or software adoption.
• Set an alert on MU and the HBM/memory complex: if memory pricing stays tight, NVDA’s demand remains intact but deployment timing risk rises; that is the key watch item before adding size to semis.