Texas Instruments vs. Cerebras Systems: Which Chip Stock Has the Edge?

Texas Instruments reported Q2 2026 revenue growth of 23% to $5.46B and a 52% EPS increase to $2.14, with gross margin rising 350bps to 61.4%; midpoint Q3 guidance implies roughly 25% revenue growth and 62% EPS growth. Cerebras grew core revenue 103% to $209.9M and holds $25.4B of RPO, but posted a $450.5M GAAP net loss, faces declining margins and significant customer concentration. The article favors TXN for its diversified growth, profitability, $6.5B trailing-12-month cash generation, $5.8B shareholder returns, and lower 10.48x forward sales valuation versus CBRS at 21.40x.

Analysis

TXN’s setup is more levered to an early-cycle industrial/auto semiconductor recovery than its AI label suggests. Incremental utilization of its 300mm fabs should drive operating leverage disproportionately over the next 1-3 quarters, while the current capital-return program limits downside if the recovery proves uneven. This is unfavorable on relative positioning for analog peers with less manufacturing cost advantage or weaker auto/industrial exposure, notably MCHP and ON; ADI is the cleaner quality peer but likely offers less earnings torque.

CBRS is effectively underwriting a concentrated, capital-intensive infrastructure build rather than simply monetizing inference silicon. The key issue is conversion quality: RPO has little valuation value until delivery milestones, cancellation provisions, payment terms, and associated capex are known. Margin dilution while capacity is rented or ramped creates a negative operating-leverage window through 2027; any delay in power interconnection, customer deployment, or wafer supply could force external financing and compress a premium revenue multiple rapidly.

The non-consensus implication is that CBRS’s architecture is not yet a material threat to NVDA or HBM/advanced-packaging demand; its avoidance of those bottlenecks is strategically useful only if it can secure enough power, systems integration, and customer software adoption to scale. Conversely, successful deployment would be incrementally positive for data-center power and thermal vendors such as VRT and power generators such as CEG, because inference capacity is ultimately constrained by energized megawatts, not only chip availability. Near-term, TXN’s reported recovery is more readily monetizable; CBRS remains a 6-18 month execution option with asymmetric downside if one major customer reduces commitments.

Key Decisions for Investors

• Initiate a 3-6 month pair: long TXN / short CBRS, sized beta-neutral. The thesis is earnings and free-cash-flow visibility versus a high-multiple business facing margin dilution and customer-concentration risk; target 15-20% relative outperformance. Exit if CBRS demonstrates two consecutive quarters of gross-margin expansion with material RPO conversion and no incremental funding need.
• Add TXN on broad semiconductor pullbacks rather than chase the next guidance print; use a 6-12 month horizon. Re-rate potential rests on sustained industrial bookings and fab utilization, while falsification is a sequential order slowdown or gross-margin reversal that indicates the recovery is inventory-led rather than demand-led.
• Use MCHP as the more tactical short hedge against a softer industrial recovery, paired with TXN rather than broad SOX exposure. TXN’s manufacturing-cost position should preserve margins better if industrial demand moderates; cover the hedge if MCHP guides bookings materially above consensus or channel inventory normalizes faster than expected.
• Maintain a watch item, not a position, in VRT and CEG for evidence that CBRS capacity converts into funded data-center projects. Require disclosed energized-MW milestones and contracted power economics before attributing meaningful revenue; a delayed interconnection schedule would negate the spillover thesis.