Zacks Industry Outlook NVIDIA, Texas, STMicroelectronics and Intel

Zacks forecasts an exceptional 2026 semiconductor cycle, citing WSTS global sales growth of 89.9% to $1.51 trillion, IDC revenue growth of 52.8%, and Gartner growth of 92% to $1.56 trillion, led by AI infrastructure and tight DRAM/HBM supply. It favors NVIDIA and Texas Instruments: NVIDIA's fiscal Q2 2027 revenue rose 106% year over year and data-center revenue increased 117% to $89 billion, while TI is expected to grow 2026 revenue and earnings by 22.7% and 55.1%, respectively. Semiconductor industry earnings estimates have risen 51.4% for 2026 and 108.8% for 2027, while the group trades at 18.32x forward earnings, a 7.7% discount to the S&P 500; risks include memory-cost pressure, tariffs, export restrictions, and geopolitical supply-chain disruption.

Analysis

The investable read-through is narrower than the broad semiconductor optimism implies: constrained HBM, advanced packaging and leading-edge wafer capacity transfer economic rents upstream to TSM and memory suppliers, while limiting unit growth for system vendors. NVDA can preserve revenue through allocation and pricing, but its incremental earnings sensitivity is now more exposed to component-cost inflation than demand; a 300-400bp gross-margin compression matters more for the multiple than another headline backlog claim. Over the next 1-3 months, the key catalyst is evidence that Blackwell/Rubin system pricing offsets HBM and packaging costs rather than merely sustaining revenue growth.

TXN is a differentiated cyclical recovery trade, not a clean AI proxy. Its internal 300mm network creates a substantial 6-18 month structural cost advantage once utilization rises, but it also embeds operating leverage in the wrong direction if industrial or auto orders soften: fixed fab costs and new capacity can turn modest revenue misses into outsized gross-margin misses. The more attractive second-order beneficiary of domestic capacity buildout is equipment/service exposure (AMAT, LRCX, KLAC), where customer capex is recognized before end-market utilization is proven.

Consensus is likely underestimating the split between AI-linked scarcity beneficiaries and the broad analog/MCU complex. Tight memory can delay automotive and industrial production schedules, impairing STM and portions of TXN even while aggregate semiconductor revenue accelerates. A Taiwan-risk shock would initially re-rate TSM and NVDA downward despite their strategic importance, while accelerating multi-year capex toward INTC; this is a hedgeable geopolitical convexity, not a near-term base-case catalyst.

Key Decisions for Investors

• Maintain an overweight long TSM over NVDA for the next 3-6 months: TSM captures AI volume and pricing across multiple accelerator customers, whereas NVDA bears greater component-cost and customer-concentration risk. Reassess if TSM advanced-node utilization or CoWoS capacity guidance weakens; target a 10-15% relative return with a 7% relative stop.
• Express AI-margin risk with a 3-6 month NVDA/TSM pair: long TSM and short a smaller NVDA notional after strength into earnings. The thesis fails if NVDA demonstrates gross-margin stabilization above 73% while sustaining sequential data-center growth above 15%.
• Initiate a 6-12 month long TXN / short STM pair on broad industrial order improvement, not before it. TXN's direct-sales model and domestic manufacturing economics should produce stronger incremental margins in recovery; invalidate if TXN book-to-bill remains below 1.0 or China revenue deteriorates materially.
• Add AMAT, LRCX, or KLAC on semiconductor-capex pullbacks rather than chasing chip designers; they monetize geographic fab duplication and capacity upgrades regardless of which accelerator vendor wins. Use quarterly foundry and memory capex guidance as the trigger; reduce if 2027 capex plans are cut by more than 10%.
• Do not add unhedged INTC exposure on policy optionality alone. Upgrade only if foundry external-customer commitments and leading-node yield disclosures improve; absent those data, subsidy support does not resolve execution and utilization risk.