Nvidia Stock Lags the 2026 Chip Rally: Why

The company at the center of the AI buildout has become the odd one out in its own sector’s rally. As of early July 2026, NVIDIA stock is up roughly 3.2% year to date — a striking laggard in a group where money has flooded into almost everything else. The VanEck Semiconductor ETF is up about 59% on the year, and the sector’s biggest gainers have left Nvidia far behind: Micron Technology has climbed a stunning 304.62%, and Advanced Micro Devices has ripped 171.25%.

A stock returning 3% while its industry returns nearly 60% is not a disaster, but it is a puzzle — because by most fundamental measures Nvidia is stronger than ever. Resolving that puzzle is the story of the 2026 chip trade.

The scoreboard

The divergence is stark when the leaders are lined up:

• Micron Technology: up ~304% year to date
• Advanced Micro Devices: up ~171% year to date
• VanEck Semiconductor ETF: up ~59% year to date
• NVIDIA: up ~3.2% year to date

Nvidia still owns roughly 81% of the AI accelerator market. It is not losing its core business. But the market prices stocks on the change in expectations, not their level — and on that measure, the names that surprised to the upside were memory and custom silicon, not the incumbent everyone already owned.

Why memory ran away with it

Memory chips have been the sector’s standout winners, and Micron is the clearest example. The company’s most recent quarter saw revenue explode 345.7% year over year to $41.46 billion, and management pointed to multi-year Strategic Customer Agreements designed to lock in demand durability rather than ride the usual boom-bust memory cycle. Most eye-catching of all, Micron guided to adjusted gross margins of about 86% for the coming quarter — above the peak gross margins Nvidia itself ever achieved.

That last figure captures why capital rotated. High-bandwidth memory sits directly in the path of every AI accelerator, and its pricing has been on a tear driven by the same forces powering the AI memory supercycle. When a commodity supplier starts posting software-like margins, investors reprice it hard — and Micron’s 304% run is the result. The tightness in supply that made those margins possible is the same story we’ve tracked in the HBM4 supply race, and it explains why Micron’s stock keeps swinging with such force in both directions.

Custom ASICs are eating the narrative

The second pressure on Nvidia’s stock is strategic, not cyclical. Broadcom’s custom AI chips — the ASICs it designs with hyperscalers like Alphabet and Meta — have become the market’s preferred way to express the “AI compute keeps growing” thesis without owning Nvidia directly.

The numbers behind that shift are large. Bloomberg Intelligence is forecasting a 27% compound annual growth rate for custom ASICs through 2033, versus 16% for merchant AI accelerators — the category Nvidia dominates. Broadcom alone is reported to carry a $73 billion AI backlog and is targeting $100 billion in annual AI chip revenue by 2027.

The logic is straightforward for the biggest buyers. A hyperscaler running its own models at enormous scale can design a chip tuned to exactly its workloads, avoid Nvidia’s margin, and reduce its dependence on a single supplier. That is precisely the trade-off between general-purpose and specialized silicon; for a primer on why the GPU became the default AI workhorse in the first place — and why purpose-built parts can undercut it — the architecture is the whole story. Every dollar of custom-ASIC growth is a dollar the market no longer automatically assigns to Nvidia’s future, even if Nvidia’s absolute revenue keeps climbing.

The other pressures on the story

Two more factors round out the picture. First, positioning: Nvidia has been the single most crowded stock in the market for two years running. When nearly everyone already holds a name, there are fewer marginal buyers left to push it higher, and the fresh money looks for the next chapter — which in 2026 was memory and custom silicon.

Second, competition at the top end. AMD’s 171% run reflects genuine traction against Nvidia in data-center accelerators, a fight we broke down in our comparison of AMD’s MI400 versus Nvidia. Nvidia’s answer is its next-generation platform, covered in our look at the Nvidia Rubin platform — but a strong roadmap defends the franchise; it does not, on its own, re-rate a stock the whole market already owns.

None of this is happening in a calm tape. The sector just came through a sharp chip-stock sell-off in early July, when the memory and equipment names that had led the rally fell hardest — a reminder that the stocks with the biggest gains also carry the highest beta when sentiment turns.

What it means

The paradox is real: Nvidia’s business has arguably never been healthier, yet its stock has been dead money in a roaring sector. The resolution is that the market spent 2026 repricing the rest of the AI supply chain — memory vendors posting record margins, custom-silicon designers winning hyperscaler commitments — while Nvidia’s dominance was already fully in the price.

Who wins. Memory makers (Micron above all) and custom-ASIC designers (Broadcom) captured the incremental enthusiasm, because they represented new information. AMD won share and a re-rating at once. The hyperscalers designing their own chips win leverage over their largest supplier.

Who is pressured. Nvidia’s stock, not its business — an important distinction. The company still commands ~81% of the accelerator market and sets the pace on performance. But the narrative that “AI compute grows, therefore buy Nvidia” has fractured into “AI compute grows, and here are five other ways to own it.” A dominant incumbent whose growth is already priced can lag for a long time even while executing well.

What to watch next. Three things. First, whether custom-ASIC forecasts translate into deployed volume or stay projections — the 27% CAGR is a bet, not yet a fact. Second, Nvidia’s own guidance in the coming earnings cycle: a genuine upside surprise on data-center demand could snap the stock out of its range. Third, the memory cycle, which has driven the sector’s biggest gains and would drive its biggest reversal if pricing rolls over. For now, the message of the 2026 chip rally is that in a maturing AI trade, dominance and stock performance are no longer the same thing.