Applied Materials, Inc. (AMAT) Presents at Goldman Sachs Communacopia + Technology Conference 2026 Transcript

Applied Materials CEO Gary Dickerson said AI data-center demand is now the company’s dominant growth driver, with leading-edge foundry data-center wafer starts expected to rise from parity with smartphones today to roughly 2:1 within a couple of years. He said 80% of wafer-fab-equipment growth in 2026 is concentrated in leading foundry/logic, DRAM including high-bandwidth memory, and advanced packaging, with a similar mix expected in 2027. The comments indicate sustained semiconductor-capex visibility in AI-related manufacturing segments.

Analysis

The important read-through is not simply higher semiconductor capex, but a sustained mix shift toward process-intensity-heavy layers where AMAT's deposition, materials engineering and packaging content can outgrow wafer-start growth. HBM stacks and advanced logic require more deposition, etch, inspection and interconnect steps per unit of compute; this favors AMAT alongside LRCX and KLAC, while the relative benefit to memory suppliers depends on whether HBM pricing remains sufficiently tight to fund capacity additions. The strongest second-order beneficiary is likely the advanced-packaging ecosystem, including TSM, ASE and AMKR, as packaging becomes a bottleneck rather than a low-value back-end service.

The market should distinguish visibility from irrevocable orders. Foundry and memory customers can preserve long-range equipment forecasts while moving delivery schedules if AI cluster utilization, hyperscaler capex, or HBM yields disappoint; that would first hit shipment timing and working-capital conversion rather than necessarily cancel the multiyear demand thesis. Over the next 1-3 months, the relevant catalyst is whether AMAT's next results convert this narrative into backlog, revenue-growth and gross-margin guidance above consensus; over 6-18 months, export-control expansion and China demand normalization remain the principal multiple risks.

Contrarianly, AMAT may be a cleaner way to express AI infrastructure than GPU suppliers if the market begins to question accelerator pricing power: leading-edge process complexity still rises even if compute demand moderates from exceptional levels. But this is not sufficient evidence to chase a post-conference move. A relative long is more attractive on weakness or after independently verifiable guidance, since equipment stocks typically discount capacity plans well before fabs receive tools.

Key Decisions for Investors

• Maintain or initiate a 6-12 month long AMAT / short SOXX pair on a 1-2 standard-deviation relative pullback: the thesis is process-intensity and packaging-content outgrowth versus broad semiconductor beta. Exit if AMAT's next revenue guide falls below consensus or if management cites material customer delivery deferrals.
• Prefer AMAT over MU for AI-memory exposure over the next 6 months: AMAT captures HBM capacity investment without direct exposure to a reversal in HBM ASPs. Reassess if DRAM suppliers signal capex discipline despite strong HBM demand, which would weaken the equipment conversion mechanism.
• Use KLAC as a confirmation watch item rather than a second outright long: sustained inspection demand and stable lead times would validate that advanced-node/HBM builds are moving from announced plans to production execution. A weak KLA order or services outlook would be an early warning for AMAT estimates.
• Do not add directional exposure solely on conference commentary; set an alert for the next AMAT backlog, deferred-revenue and China-revenue disclosures. Evidence of backlog growth with stable gross margin supports adding, while a China-related mix headwind or margin dilution argues for reducing the position.