Chip "first choice" Nvidia, replacing memory; Morgan Stanley's reasoning: the stock price has lagged for a long time, and the "growth peaking in 2026" theory has been disproven.
```
NVIDIA’s stock price has barely moved over the past two quarters, but Morgan Stanley’s assessment is: the fundamentals are actually getting stronger, and the market is stuck on two issues—whether growth can “endure” through 2026, and whether market share will gradually be eroded by ASICs and AMD. Morgan Stanley has reinstated NVIDIA as its “top pick” in the semiconductor sector, maintains its Overweight rating and $260 price target, and views the current valuation as a rare entry window.
According to Wind Chasing Trading Desk, Morgan Stanley North America semiconductor analyst Joseph Moore states in his latest report that NVIDIA’s 2027 earnings are priced at only about 18 times PE, which is a surprisingly good entry point. His logic is not to bet on another beat next quarter, but to bet that investors’ doubts about “growth durability” will start to ease in the coming months.
The evidence supporting this view comes mostly from the supply chain and customer behavior: the report mentions that hyperscale cloud providers are locking in upstream orders for longer cycles, and there are cases of “three-year orders, some with full prepayment.” Actions like these are hard to reconcile with “planning to hit the brakes next year.” However, the report also admits that downstream cash flow pressure is a real problem: some cloud infrastructure businesses are in a negative cash flow state, and extrapolating from financials, next year’s capital expenditure may indeed slow down.
As for market share, Morgan Stanley does not deny that NVIDIA may see a slight dip in 2026: it estimates NVIDIA’s revenue share at about 85%, ASIC slightly above 10%, and AMD slightly below 5%. Given that NVIDIA’s scale is already approaching a “$80 billion per quarter” constraint, it’s not surprising that peers grow slightly faster and NVIDIA gives up 1–2 percentage points in share.
The essence of two-quarter stagnation: the market is betting “growth won’t last,” not denying current prosperity
Morgan Stanley attributes NVIDIA’s recent “lack of momentum” to the structure of expectations: the market already assumes things are strong now, but the real divergence is about 2027.
They mention a strange comparison—after shifting top pick from NVIDIA to SanDisk, then to Micron, memory stocks cumulatively rose 300%–900%, while NVIDIA stock “stood still”; meanwhile, NVIDIA’s “current quarter” earnings outlook rose 38% over 6 months. This means it’s not the data suppressing the share price, but doubts about the cycle's longevity.
The report also places these emotional swings in a historical pattern: for each of the past three years, markets have been uneasy about “the next year” at the start of the year, until visibility improves, and then the stock outperforms in stages. It bets 2026 will repeat this process.
Hard signals from the supply chain: Three-year locked orders, even full prepayments
Morgan Stanley believes durability is harder to prove, but the upstream behavior they see is increasingly extreme. The report notes hyperscale customers place three-year orders with memory vendors, sometimes with full prepayment; an extreme case is “this quarter already receiving 100% of 2028 revenues,” at volumes several times current levels.
They view these prepayments as “durability clues”: if customers really intend to slow next year, it’s hard to explain locking in future supplies with cash now. The report emphasizes this is just one of many signals—supply chain feedback shows multiple links are preparing for sustained growth.
Counter-evidence downstream comes from cash flow: Some cloud infrastructure businesses still burning cash
Morgan Stanley does not gloss over the “cash flow problem.” The report mentions that some cloud infrastructure businesses are in negative cash flow: this can of course continue (balance sheets are strong, financing is available), but purely from financial numbers, next year’s spending does “look set to slow.”
The explanatory framework is more akin to a “time lag”: NVIDIA’s management responded on calls that cloud vendors will earn much more from these investments than market models project;
The report adds that cloud GPU margins are high, and market participants discuss how “the compute supply gap expands by single digits daily.” In other words, short-term cash flows may look ugly, but that doesn’t mean investment returns aren’t real.
Share may slip slightly, but clients’ focus has shifted to Rubin
On competitive landscape, Morgan Stanley offers a clear breakdown: NVIDIA revenue share about 85%, ASIC slightly above 10%, AMD slightly below 5%. It expects peer growth to be a bit faster in 2026, with NVIDIA’s share slipping 1–2 points, due to a “physical” reason: the larger the scale, the greater the supply constraint, the harder to sustain the same growth rate.
But the report contends clients are truly fixated on Rubin (expected to ship in the second half), while prior competition is mainly Blackwell vs. various alternatives. They admit the moat has “been a little eroded,” since leading model developers are more architecture-neutral, simultaneously using NVIDIA, AMD, Google TPU and in-house ASICs.
Still, their research finds that even in areas where third parties stress “lower total ownership cost,” customer preference often tilts to NVIDIA. The report notes the two largest ASIC users and the two top potential AMD users may still let their business growth with NVIDIA exceed 80% in 2026.
Furthermore, for more aggressive bull narratives (advanced model developers suggest needing “hundreds of GW compute” by 2029, vs. about 30GW this year), Morgan Stanley is cautious: it doesn’t buy that number, but also makes clear it sees no signs of the “cycle ending” in 2026.
The key constraint is capacity: semiconductor capacity cannot ramp up supply to that scale quickly. More importantly, global supply chain feedback is not “orders weakening,” but “using cash to lock in growth beyond 2028 early.”
For the upcoming GTC conference, Morgan Stanley expects it may not fully convince skeptics on the sustainability of capex, but will likely clarify the market share debate. This showcase will be similar to 2024—presenting a more complete four-year roadmap and emphasizing that competition is not just about chips, but also cabinets and ecosystem building; at the same time, it expects Groq IP will have a place in the roadmap.
Risk Warning and DisclaimerThe market involves risk, invest cautiously. This article does not constitute personal investment advice, nor does it take into account any individual user’s special investment objectives, financial situation, or needs. Users should consider whether any opinions, views, or conclusions expressed herein are suitable to their own situation. Invest accordingly at your own risk. ```