Texas Instruments: CapEx peak unlocks a multi-year margin and cash flow re-rating

Stevie AI on Texas Instruments Incorporated (TXN-USA | texasinstrum)

3/30/2026

Summary

Texas Instruments is the world's largest analog semiconductor company, designing and manufacturing chips that sit invisibly inside virtually every piece of industrial equipment, automobile, and electronic device on the planet. The structural insight powering this thesis is simple but underappreciated: TXN spent four years absorbing a brutal CapEx cycle — peaking at $4.55B in 2025 — to build 300mm fab capacity that competitors cannot replicate at scale. That investment cycle is now cresting. As depreciation stabilizes and utilization rates climb through an industrial and automotive end-market recovery, TXN is positioned to deliver an earnings inflection that the market, still anchored to trough-cycle margins and depressed 2024 EPS of $5.20, has not yet fully priced in. The financial trajectory supports this view. Revenue trough in FY2023-2024 at $15.6B, reflecting the sharpest inventory correction in the analog industry in over a decade, as customers in industrial and automotive channels destocked aggressively. FY2025 showed early-cycle recovery, with revenue recovering to $17.7B and EPS edging up to $5.45 — still well below the company's normalized earnings power. These numbers matter because the trough itself validates the thesis: TXN absorbed peak depreciation, funded a $4.6B annual CapEx program, and still generated positive cash flows, demonstrating the resilience of its business model. The recovery from here is not speculative; it is a function of utilization rates on capacity that already exists. We initiate coverage with a BUY rating and a 12-month price target of $195, rising to $344 by FY2029, based on a 30x forward P/E applied to our EPS estimates. We apply 30x — a modest premium to the S&P 500 but a discount to high-growth semiconductor peers — to reflect TXN's combination of above-market EPS growth (18%+ CAGR from 2025 to 2029), durable competitive moat via 300mm manufacturing, secular analog content tailwinds in AI infrastructure and EVs, and a capital return program that compounds per-share value. At the current price of $190.33, shares trade at just 34.9x depressed FY2025 EPS, but only 29.2x our FY2026 estimate of $6.52 — a multiple that prices in no recovery premium whatsoever. The risk/reward is compelling.

Thesis

1. **The 300mm Manufacturing Moat Is a Structural Advantage That Takes a Decade to Replicate** Texas Instruments' decision to aggressively convert its manufacturing base to 300mm wafers is the single most important strategic bet in the analog semiconductor industry over the past decade — and it is now paying off. A 300mm wafer contains roughly 2.4x the die area of a 200mm wafer at comparable cost per wafer start, translating to a 40%+ cost-per-chip structural advantage over competitors still running predominantly 200mm fabs. This is not a temporary lead; it took TXN itself nearly a decade and roughly $15B in cumulative CapEx to build this position. Analog Devices, ON Semiconductor, Microchip Technology, and the broader analog competitive set cannot close this gap in a single investment cycle. The strategic significance extends beyond unit economics. The capacity built — including the Sherman, Texas and Lehigh Valley, Pennsylvania 300mm fabs — is being partially funded by CHIPS Act investment tax credits and government assistance estimated at $350-500M annually through the forecast period. This represents genuine incremental P&L and cash flow support that competitors relying on Asian foundry capacity cannot access. The net effect: TXN is building a cost and supply-chain security advantage simultaneously, supported by government subsidy, at a time when customers in industrial and automotive end markets are actively seeking domestic supply chain resilience post-COVID. For investors, the moat is not abstract. It shows up in pricing power — TXN has historically maintained gross margins in the 60-65% range through mid-cycle conditions — and in customer stickiness, as analog components are designed into end products for five to ten year lifecycles with extremely high switching costs. Once a TXN analog chip is designed into an industrial power supply or an automotive ADAS module, the probability of displacement at the next model refresh is low. This recurring, sticky revenue base with cost advantages compounding through scale is the foundation of the investment thesis. 2. **Industrial and Automotive Recovery Is a Cycle Turn, Not a Hope** The FY2023-2024 revenue decline was driven almost entirely by a customer-level inventory correction in industrial (33% of 2025 revenue, $5.8B) and automotive (33%, $5.8B) end markets — not by secular demand destruction. Industrial customers, which had overbought analog components during the post-COVID supply crunch, spent 18-24 months burning down excess inventory rather than placing new orders. Automotive OEMs and Tier 1 suppliers followed a similar pattern, amplified by EV demand uncertainty in 2023-2024. The result was a demand air pocket that compressed TXN's revenue by roughly 20% from peak to trough and pushed gross margins below 55% as fixed depreciation costs spread over lower volumes. The turn is already in motion. FY2025 revenue of $17.7B represents 13.5% recovery from the trough, and management's Q1 FY2026 guidance of $4.32B-$4.68B (midpoint $4.5B) implies sequential growth above seasonal norms — a signal that order momentum is genuine rather than episodic. Our forecast builds to $19.1B in FY2026 (+8% YoY) and $21.0B in FY2027 (+10% YoY), assuming continued but measured normalization. These are not heroic assumptions; $25.9B in FY2029 revenue represents a compound annual growth rate of roughly 10% from FY2025 — consistent with TXN's own long-term financial model and below the growth rates embedded in prior cycle peaks. The automotive vector deserves particular emphasis. Electric vehicles carry 2-3x the analog semiconductor content of internal combustion engine vehicles — in power management, battery management systems, thermal management, and ADAS — and the penetration trajectory remains intact despite near-term EV demand volatility. TXN's $5.8B automotive revenue base in FY2025 is well-positioned to compound above overall company growth as the EV transition accelerates through FY2027-2029. This is a structural demand tailwind layered on top of a cyclical inventory recovery. 3. **CapEx Step-Down Is the Earnings Catalyst the Market Is Underpricing** The most important near-term financial dynamic for TXN is the transition from peak CapEx to normalized CapEx — and its cascade effect on free cash flow and earnings per share. At $4.55B of CapEx in FY2025, TXN was investing at nearly 26% of revenue, an extraordinary intensity level for a mature analog manufacturer. This investment funded real assets — 300mm fab capacity that will serve the company for 20-30 years — but it also created a depreciation overhang that suppressed reported gross margins to the 55-57% range, well below the company's 60%+ normalized target. Starting in FY2026, CapEx steps down sharply to approximately $2.5B (~13% of revenue), then rises modestly to $2.8B-$3.2B in FY2027-2029 as utilization-driven maintenance and incremental tooling begins. This reduction of roughly $2B in annual capital expenditure flows almost entirely into free cash flow, which we forecast recovering from a depressed level to $5.3B in FY2026, $6.4B in FY2027, $7.8B in FY2028, and $9.0B in FY2029. Critically, depreciation — which lags CapEx and continues rising for 12-18 months after CapEx peaks — stabilizes as a percent of revenue by mid-FY2026, allowing gross margins to begin the recovery path toward 60%+ by FY2028-2029. The earnings math is compelling. Our EPS forecast goes from $5.45 in FY2025 to $6.52 in FY2026, $7.87 in FY2027, $9.68 in FY2028, and $11.46 in FY2029 — a four-year CAGR of approximately 20%. This is not multiple expansion; it is earnings growth driven by three simultaneous tailwinds: revenue recovery (volume leverage on existing capacity), gross margin expansion (depreciation stabilization plus utilization improvement), and per-share accretion from buybacks. The market at $190 is pricing TXN as if these tailwinds either do not exist or do not persist. They do both. 4. **Data Center Is a Secular Upside Option That Is Not Fully Priced In** Data center revenue, estimated at 9% of FY2025 revenue (~$1.5B), is TXN's fastest-growing end market and represents a genuinely underappreciated optionality in the investment thesis. The AI infrastructure build-out — driven by hyperscaler investment in GPU clusters, custom ASICs, and the supporting power and cooling infrastructure — generates substantial demand for analog semiconductors. Every AI server rack requires power management ICs, voltage regulators, temperature sensors, and interface logic that are precisely TXN's product categories. This is not a niche exposure; it is the analog content of the AI capex supercycle. Our forecast assumes data center revenue grows from 9% of total revenue in FY2025 to approximately 13% by FY2029 — implying roughly $3.4B of data center revenue at FY2029 scale, more than doubling in four years. This growth rate is actually conservative relative to hyperscaler CapEx trends. Microsoft, Google, Amazon, and Meta have each guided to materially higher infrastructure spending in FY2026-2027, and the content per rack in AI-optimized infrastructure is 3-5x that of traditional compute. If data center penetration reaches 15-16% of revenue rather than our 13% base case, the revenue and margin upside to our FY2028-2029 estimates is meaningful. Importantly, data center revenue carries above-corporate-average margins given the relative complexity and mission-criticality of the applications. Power integrity in an AI training cluster is not a commodity problem; it requires validated, high-reliability components with proven field performance. TXN's reputation and catalog depth in power management gives it a credible claim to this revenue that newer entrants cannot easily displace. The segment is therefore not only a top-line growth driver but a structural mix tailwind for gross and operating margins. 5. **Capital Allocation Discipline Compounds Per-Share Value Through the Cycle** Texas Instruments has returned more than $40B to shareholders through dividends and buybacks over the past decade, establishing one of the most consistent capital return programs in the semiconductor sector. Going forward, the combination of a recovering dividend coverage ratio and an improving FCF trajectory creates the conditions for per-share value compounding that investors are underestimating at current prices. Our forecast assumes approximately $1.5-2B of annual share repurchases, sufficient to drive a gradual decline in diluted share count and resulting per-share accretion ahead of absolute net income growth. This is conservative relative to historical buyback intensity; as net debt stabilizes (peaking at ~$14.5B in FY2028 before declining in FY2029) and FCF accelerates toward $9B, management has the capacity to accelerate buybacks without compromising the balance sheet or the dividend. The dividend itself is expected to grow at 5-6% annually — a payout that, at the current stock price, yields approximately 2.5%, providing income support during any near-term volatility. The February 24, 2026 Capital Management Call is a near-term catalyst for clarifying the capital allocation framework and potentially signaling an inflection in buyback cadence. If management uses this forum to articulate FCF-per-share targets or updated return-of-capital commitments — as they have historically done at similar investor events — it could serve as the catalyst that bridges the market's current anchor on trough earnings to a forward-looking recovery multiple. 6. **The Market Is Still Anchoring to Trough Economics — the Mispricing Is Temporal, Not Structural** At $190.33, TXN trades at 34.9x FY2025 EPS of $5.45, which on the surface appears expensive. But this multiple is applied to cyclically depressed earnings that do not reflect the company's normalized earnings power. On our FY2026 estimate of $6.52, the stock trades at 29.2x — and on FY2027's $7.87, just 24.2x. These multiples are not demanding for a company with a defensible moat, 18-20% EPS growth, rising FCF, government-supported domestic manufacturing, and secular tailwinds in automotive electrification and AI infrastructure. The mispricing is temporal: the market is using the wrong denominator. Investors applying trailing or near-term multiples to trough EPS are conflating cycle trough with structural deterioration. TXN's competitive position has not weakened during the inventory correction — if anything, it has strengthened, as competitors without 300mm capacity are less able to offer the cost, availability, and supply chain security that customers value most in the early stages of a restocking cycle. A 30x multiple applied to our FY2026 EPS of $6.52 implies a price target of $195.60 — approximately in line with current trading, which means the market is awarding precisely zero recovery premium. As EPS grows to $7.87 in FY2027 at the same 30x multiple, the implied price target rises to $236. By FY2029 at $11.46 EPS, the price target reaches $343.80. The compounding effect of EPS growth alone — without any multiple expansion — generates 80%+ total return from current levels over four years. Add dividend income of ~2.5% annually and the total return picture is highly compelling for a large-cap quality compounder.

Risks

1. **Depreciation Overhang Could Persist Longer Than Forecasted** The single largest risk to the investment thesis is that depreciation from the 300mm fab buildout continues to burden gross margins beyond FY2026, either because utilization rates ramp more slowly than expected or because additional tooling and capacity decisions extend the depreciation timeline. Gross margin declined 150 basis points sequentially in Q4 2025, and management acknowledged that the depreciation headwind does not reverse immediately when CapEx steps down — it lags by 12-24 months as existing asset bases continue to amortize. If utilization rates remain below 70-75% due to slower-than-expected industrial or automotive recovery, gross margins could stay trapped in the 55-58% range through FY2027, compressing EPS materially below our estimates. Every 100 basis points of gross margin underperformance relative to our 60%+ FY2028-2029 forecast represents approximately $235M of net income and roughly $0.27 of EPS dilution. 2. **Industrial and Automotive End Markets Could Disappoint on Recovery Timing** Our thesis depends critically on industrial and automotive end markets recovering through FY2026-2027 at a pace consistent with normal post-correction inventory normalization. If macro conditions deteriorate — particularly in Europe, which is a major destination for TXN's industrial revenue — or if automotive production volumes decline due to EV demand softness, trade disruptions, or OEM production cuts, the volume recovery underpinning our gross margin and EPS forecasts could be delayed by 12-18 months. A 5% revenue miss relative to our FY2026 forecast of $19.1B would reduce EPS by approximately $0.40-0.50, and would likely also compress the market multiple as investors extend their recovery timeline assumptions. The inventory correction of 2023-2024 was deeper and longer than most forecasters anticipated; the recovery could prove similarly uneven. 3. **Net Debt Position Limits Financial Flexibility and Creates Refinancing Exposure** TXN carries a substantial and rising net debt load, growing from $12.4B in FY2026 to a peak of $14.5B in FY2028 before modestly declining in FY2029. This leverage was deliberately taken on to fund the 300mm buildout, but it creates real constraints. Interest expense is a meaningful drag on net income, and the refinancing of debt maturities in a higher-for-longer interest rate environment could increase interest costs by $100-200M annually relative to the rates at which the existing debt was issued. More importantly, elevated net debt limits the pace of buybacks and creates covenant or credit rating sensitivity if earnings undershoot expectations. If TXN faces a revenue or margin shortfall while managing a $14B+ net debt position, the margin of safety for capital returns narrows meaningfully, and the dividend — while likely protected — could face coverage questions. 4. **Competitive Pressure From Domestic and International Analog Players** While TXN's 300mm manufacturing advantage is real, it is not impenetrable. Analog Devices, ON Semiconductor, STMicroelectronics, Renesas, and Infineon are each investing in capacity, portfolio expansion, and customer relationships in the same industrial and automotive end markets. In China specifically, domestically backed analog semiconductor companies — including SMICS analog design spin-offs and funded startups — are beginning to qualify into mid-tier industrial applications, potentially displacing TXN at the lower end of its catalog in a key growth geography. If pricing pressure in commoditized analog categories intensifies — particularly in 8-bit and 16-bit embedded processing, where competition is most acute — it could limit TXN's pricing power and compress the gross margin recovery trajectory embedded in our FY2027-2029 estimates. 5. **CHIPS Act and Government Support Is Politically and Legislatively Contingent** Our forecast includes $350-500M annually of CHIPS Act investment tax credits and government assistance as a meaningful P&L and cash flow support item. This support is not guaranteed beyond the current legislative framework. Changes in U.S. trade policy, budget reconciliation processes, or administration priorities could delay, reduce, or eliminate the subsidy component of the CHIPS Act benefits. Additionally, compliance requirements — including restrictions on expanding capacity in certain geographies, executive compensation limits, and profit-sharing provisions for projects exceeding return thresholds — could create operational friction or limit TXN's strategic flexibility. If government support is reduced by 50% relative to our forecast, the impact to FCF is approximately $175-250M annually — modest in isolation, but meaningful in the context of a thesis that depends on FCF recovery as a valuation catalyst. 6. **AI Data Center Demand Could Be Cyclical Rather Than Structural** Our forecast assumes data center revenue grows from 9% to approximately 13% of total revenue by FY2029, driven by sustained hyperscaler AI infrastructure investment. However, if AI capital expenditure by hyperscalers decelerates — due to regulatory pressure, ROI disappointment on AI applications, or technology substitution that reduces analog content per compute unit — the data center revenue growth embedded in our model could reverse. Hyperscaler CapEx is itself a capital cycle, and the analog content tailwind from AI infrastructure, while real today, could plateau or contract if the build-out phase of AI infrastructure gives way to an optimization phase with lower hardware intensity. TXN does not break out data center revenue separately, making this exposure difficult to monitor in real time; investors should treat the data center upside as a call option on sustained AI CapEx rather than a base-case certainty.

📈 Price Targets