Prediction: 2 AI Semiconductor Stocks That Will Join Micron and Broadcom in the $1 Trillion Club by 2028

AMD and ASML are presented as potential $1 trillion semiconductor companies by 2028, supported by sustained AI infrastructure spending and sharply expanding earnings capacity. AMD's data-center revenue rose 107% year over year last quarter, with AI data-center revenue expected to grow well above 100% in the current quarter; consensus EPS is projected to rise from $2.65 last year to $15.61 next year. ASML plans to expand manufacturing capacity 30% next year and potentially another 30% in 2028, a cumulative 69% increase that could support roughly €74 billion of 2028 revenue versus its prior €60 billion 2030 target.

Analysis

AMD’s rack-level push changes the relevant debate from accelerator share to system-level attach and data-center power density. The upside is not linear: successful integration can lift content per deployed rack, but it also brings lower-margin networking, memory, and support content onto AMD’s P&L, so revenue-share gains may not translate proportionately into gross-margin expansion. The near-term valuation risk is therefore execution against customer qualification schedules and the credibility of very aggressive earnings expectations, not merely demand for AI compute.

ASML is the cleaner second-order beneficiary because incremental leading-edge logic and HBM capacity ultimately require lithography intensity regardless of which accelerator vendor wins. Capacity additions should improve delivery optionality, but orders can remain lumpy as customers manage prepayments, tool lead times, and export-control exposure; the market will reward backlog quality and service revenue more than headline manufacturing capacity. A stronger ASML order book would also support adjacent wafer-fab equipment names, but AMAT and LRCX lack ASML’s monopoly economics and are more exposed to memory-cycle capex volatility.

The consensus likely underweights a potential divergence between AI infrastructure demand and AI-hardware vendor economics over the next 6-18 months. Hyperscalers can absorb more systems while pressing suppliers for rack-level pricing, creating a scenario where AMD unit growth is strong but margins and multiples disappoint; ASML’s installed-base/service model is less vulnerable to this bargaining dynamic. Conversely, a broad reduction in hyperscaler capex or tighter restrictions on China-bound tools would hit ASML’s earnings visibility faster than its long-duration narrative implies.

Key Decisions for Investors

• Prefer a 6-12 month long ASML / short AMD pair rather than outright AI-beta exposure. ASML offers monopoly-backed pricing and backlog conversion; AMD must clear product-ramp, software, and rack-margin hurdles simultaneously. Reassess if AMD demonstrates sustained AI segment gross-margin expansion or if ASML bookings weaken materially for two consecutive quarters.
• For AMD, wait for first customer deployment evidence and disclosed rack economics before adding directional exposure. A long is justified only if management quantifies revenue conversion, supply availability, and gross-margin trajectory; absent that, high earnings-expectation sensitivity creates asymmetric downside on any qualification delay over the next 1-3 months.
• Use ASML earnings/order updates as an alert for a broader equipment basket trade: add selective AMAT/LRCX exposure only if leading-edge and memory tool demand both improve, rather than buying on announced capacity alone. Falsifier: rising cancellations, delayed customer acceptances, or incremental export restrictions that reduce the addressable installed base.
• Avoid treating META’s announced deployments as a standalone supplier-revenue catalyst. The relevant watch item is whether hyperscaler capex guidance shifts toward internally designed accelerators or third-party racks; a custom-silicon pivot would pressure AMD’s achievable share while retaining much of ASML’s underlying lithography demand.