Texas Instruments Is a Worthy AI Picks-and-Shovels Play I Keep Buying

Texas Instruments reported Q2 2026 revenue of $5.463B (+22.8% YoY) and diluted EPS of $2.14 vs a $1.9356 estimate, with analog revenue up 26% and data center revenue doubling YoY. Trailing free cash flow was $6.53B (33.6% of revenue) as quarterly capex eased to $514M from $1.305B a year earlier, supported by CHIPS Act cash ($850M in Q2 2026; $1.6B incentives over the trailing 12 months). TI also declared a $1.42/share quarterly dividend (=$5.68 forward annual), continuing its streak of dividend raises through 2026 while investors remain watchful for semiconductor-cycle-driven earnings volatility.

Analysis

TXN is becoming a cleaner way to express AI infrastructure spending than the usual GPU basket: the monetization is less about unit hype and more about every incremental watt of server power needing analog conversion, protection, and control silicon. That gives TI a broader socket base and more diversified end demand than NVDA, but it also means the upside is steadier and less explosive; the market is likely still underpricing the duration of this power-management cycle while overestimating how quickly analog demand mean-reverts.

The biggest second-order winner is TI’s own capital intensity cycle: as capex rolls off while revenue inflects, incremental free cash flow can re-rate the equity even without heroic top-line growth. Against ADI, TXN screens as the higher-quality cash-return vehicle at a lower multiple, which opens room for multiple convergence if data-center and industrial demand hold for 2-4 quarters. The loser is any analog peer with weaker scale or less 300mm manufacturing leverage, because TI can defend share with both cost and supply assurance when customers want long-life, high-reliability parts.

Main risk: this is still a cyclical semiconductor complex, so a 1-2 quarter industrial or auto pause would hit reported earnings before the FCF story fully shows up. The key falsifier is not sentiment but operating data: if data-center revenue growth normalizes sharply or gross margin fails to expand despite lower capex, the thesis on FCF-per-share acceleration breaks. Over 6-18 months, the debate is whether TXN deserves a structural premium for dividend durability and domestic manufacturing, or whether the market eventually treats it as a mature cyclical and caps the multiple.

Contrarian view: the consensus may be too focused on AI as a GPU story and not enough on the power stack underneath, but it may also be too willing to extrapolate one strong quarter into a multi-year upcycle. The right stance is selective: own TXN as a quality compounder on pullbacks, but avoid paying peak-cycle multiples until industrial breadth confirms. NVDA remains the higher-beta AI expression; TXN is the lower-beta picks-and-shovels version with better income support and less narrative risk.

Key Decisions for Investors

• Long TXN vs short ADI as a 3-6 month relative-value trade: TXN has stronger free-cash-flow inflection and a lower trailing multiple; cover if TXN premium to ADI widens by ~20% or if TXN data-center growth slows materially.
• Buy TXN on weakness tied to broad semiconductor risk-off, using a 1-3 month horizon: thesis is multiple support from improving FCF/share, not near-term EPS momentum; invalidate if next two prints show margin compression despite lower capex.
• For a lower-beta AI infrastructure basket, pair long TXN with short NVDA small size: use this only if the goal is exposure to AI capex with less valuation risk; if NVDA continues to reaccelerate on GPU demand, the pair can underperform quickly.
• Watch ADI and MCHP for competitive spillover: if TI’s data-center mix keeps rising, expect pricing pressure and share loss in less differentiated analog names; use ADI weakness on any TI strength as a relative short candidate.
• Alert level: if TXN’s forward guidance implies industrial demand re-stocking rather than sustainable end-market growth, reduce exposure; the stock’s downside case is a cyclical fade, not a balance-sheet problem.