Written by: Rita
Tide Guide
Semiconductor equipment stocks have fallen 30% from their highs but are still up 80% year to date.
Bernstein provides an answer using a bottom-up calculation: for every additional 1 GW of computing power built in AI data centers, $8 billion more must be spent on equipment. At an annual addition of 50 GW, cumulative equipment spending from 2027 to 2029 will exceed $700 billion, driving annual WFE toward $300 billion.
This number carries significant weight. The current market pricing for equipment stocks implies annual WFE spending of only $120 to $150 billion—half of what’s needed. If AI construction proceeds at its current pace, equipment stocks are not just reasonably valued—they may even be undervalued.
Bernstein is overall bullish on the equipment sector, with the strongest preference for Applied Materials (AMAT). More than half of the incremental wafer demand comes from DRAM and HBM, and Applied Materials has the largest exposure in this area. Applied Materials, Lam Research, KLA, ASML, Tokyo Electron, Kokusai, and Lasertec all outperformed the market, while Screen flatlined.
$8 billion equipment bill behind 1 GW of computing power
Bernstein broke it down in detail: a single Vera Rubin rack consumes 65 wafers, covering logic, HBM, DRAM, and NAND. Outside the rack, there are additional demands for server CPUs, accompanying memory, and storage.
When summed up, for every additional 1 GW of annual computing power, 46,000 wafers per month of fabrication capacity are required. Half of this is allocated to DRAM and HBM, 20% to NAND, and 10% to advanced logic. This translates to exactly $8 billion in equipment investment.
This figure doesn't include the replacement of old equipment. A batch of legacy computing power will be upgraded between 2027 and 2029, so actual demand will be even higher.
50GW scenario: WFE annual spending set to reach $300 billion
By 2030, if an additional 50 GW of computing power is added annually—50 GW above the 2026 baseline—the cumulative WFE spending over three years will exceed $700 billion. Adding the baseline non-AI demand of approximately $120 billion per year, annual WFE spending will rise from $200 billion to nearly $300 billion.
What if we go even faster? 75 GW or 100 GW, $300 billion—these are just the starting points.
Market expectations are still far off. The consensus implies annual WFE spending of only $120 to $150 billion—more than half of what’s needed. If Bernstein’s baseline scenario materializes, there’s significant room for valuation recovery in equipment stocks.
Do the math: Are equipment stocks really expensive right now?
Bernstein ran three calculations, directly comparing current valuations.
50GW scenario. Applied Materials' EPS rises from the consensus of $18.70 to $24.30 in 2028, a 30% increase, and to $30.60 in 2029, a 60% increase. The corresponding P/E ratio declines from 23x to 15x, and further to just 11x in 2029.
75 GW scenario. EPS reaches $34.2 in 2028, an increase of over 80%. It rises to $46.7 in 2029, more than doubling. P/E ratio is 11x in 2028 and 8x in 2029.
100 GW scenario. EPS reaches $44.7 in 2028, more than doubling. It rises to $62.4 in 2029, more than tripling. P/E ratio is 8x in 2028 and 6x in 2029.
Lam Research and KLA have similar flexibility. The conclusion is straightforward: as long as AI construction continues, current equipment stocks are much cheaper than the market thinks.
Why Applied Materials?
In the incremental wafer demand, DRAM and HBM account for 55%. Applied Materials has the highest exposure in the DRAM and HBM equipment segment across the entire sector, which is Bernstein’s core reason for favoring it.
The ratings for other underlying assets remain unchanged. Lam, KLA, ASML, Tokyo Electron, Kokusai, and Lasertec all outperformed the market, while Screen matched market performance.
Tide View
The market knows that AI requires data centers, data centers require chips, and chips require equipment. But no one has carefully calculated exactly how much equipment is needed, what revenue that corresponds to, and what valuation that implies.
Bernstein has calculated that 1 GW corresponds to $8 billion in equipment, and 50 GW corresponds to $300 billion in WFE, implying a P/E ratio of 15x for equipment stocks. However, the current market pricing implies $150 billion in WFE and a P/E ratio of over 20x.
The doubling difference here represents the expectation gap.
The risks are also very real: Will the construction pace slow down? Will equipment production keep up? Will customers cancel orders during a downturn? Each is a variable. The volatility of equipment stocks has always been greater than that of semiconductors themselves.
But one thing is clear: as long as the narrative around AI computing power construction persists, the valuation of equipment stocks has not yet reached its peak. Whether the current pullback is a risk or an opportunity ultimately depends on whether investors believe in the pace of AI infrastructure development.
Disclaimer
This article is a summary and interpretation by Chaoxiang Research of a third-party brokerage research report (Bernstein, July 20, 2026). The ratings, target prices, earnings forecasts, and related judgments cited herein are the views of the brokerage’s analysts and represent only the position of their respective institution; they do not reflect the views of Chaoxiang Research nor constitute any investment advice.
The market carries risks; invest with caution. This article should not be used as a basis for buying or selling any securities. Investors should make investment decisions based on their own independent judgment.