Morgan Stanley Cuts NVIDIA FY2028 Growth Forecast to 53.5%, Maintains Overweight
nashnova research
After Jensen Huang's departure as CEO, Morgan Stanley cut Nvidia's FY2028 revenue growth forecast from 70% to 53.5% and lowered its target to $267 — calling the transition a major setback, not a structural crack.
How deep are the cuts, and where do they land?
FY2028 revenue forecast cut 9.7%, EPS cut 10.9%. FY2029 sees similar reductions. Gross margin barely moved — down just 0.09 percentage points.
The cuts land in FY2028's back half: Q3 and Q4 EPS dropped from $3.98 and $4.78 to $3.49 and $3.78. The first two quarters barely changed. This means → Morgan Stanley sees the risk as back-loaded; near-term earnings are largely intact.
Morgan Stanley itself calls these numbers "conservative" — building in a cushion first, ready to revise back up once uncertainty clears.
How far back does this revision take us?
In August, Nvidia raised its CY2027 growth framework to roughly 70%. Morgan Stanley followed, lifting FY2028 revenue from $598.8 billion to $690.2 billion.
The new forecast sits at $622.98 billion — close to the pre-August level. In plain terms = August's optimistic upgrade has been almost entirely erased; earnings expectations have temporarily retreated to where they stood before Nvidia's 70% framework.
This reflects Morgan Stanley's dual caution on the leadership change and export curbs — not a bearish call on fundamentals, but a reset to a "safe baseline."
Who takes over, and why is speed critical?
After Huang's departure, CFO Colette Kress stepped in as interim CEO. No permanent successor has been named.
Morgan Stanley sees Kress as a highly capable CFO who can keep operations running, but the "interim" tag creates friction in three areas: making long-term customer commitments, organizing large financing syndicates, and driving internal consensus on strategy.
The most-discussed internal candidate is data-center head Ian Buck — familiar to customers and supply-chain partners, but his strategic vision and willingness to take the role remain unknown. This means → the sooner a permanent CEO is named, the faster uncertainty narrows — Morgan Stanley flags this as its single most important watchpoint.
Will the supply chain actually break without Huang?
Elon Musk said publicly that many Nvidia deals "were sealed with a handshake with Jensen, not a written contract," implying the departure threatens supply-chain stability.
Morgan Stanley thinks this concern is overstated: Nvidia has locked in most capacity for the next two years through prepayments, with the majority of tight components for the next 12 months already covered.
In plain terms = the materials are bought, the capacity is locked. Handshake-deal risk is a medium-to-long-term issue, not an immediate one.
How do tighter export controls factor into the forecast?
Washington accelerated discussions this week on further tightening export restrictions. The likely focus: Chinese models renting overseas cloud compute — a practice currently permitted in most cases.
Morgan Stanley offers two buffer arguments: first, the affected compute is mainly used for inference, so if that demand disappears, other open-weight models may fill the gap; second, overall demand remains far from satisfied, and freed-up hardware would be absorbed elsewhere quickly.
Morgan Stanley stresses, however, that both buffers are assumptions — the final scope of the restrictions has not been published, so the model carries a built-in cushion. This means → if restrictions come in milder than expected, there is room to revise up.
What about the neocloud model and the CUDA-moat debate?
Nvidia's neocloud revenue-sharing arrangement — the company underwrites a minority stake, leads syndicated financing, and takes a share of cloud-provider profits — can continue under Kress, Morgan Stanley believes. But the interim tag may make it harder for smaller neoclouds to secure financing; capital is more likely to flow to the large hyperscalers.
Neoclouds are still expected to contribute close to 25% of revenue, but the mix will skew toward established players.
As for the "CUDA moat is crumbling" narrative, Morgan Stanley sees it as overblown: large frontier-model companies already use hardware-agnostic approaches. Nvidia's real moat is higher tokens-per-watt plus the full hardware-software ecosystem — not CUDA itself. From 2022 until this week, almost nobody debated this, and Nvidia's revenue share stayed above 80%.
市场有风险，内容仅供研究参考，不构成投资建议。