Texas Capital Bancshares, Inc. TCBI

Key data

The business

Texas Capital is a $33.9B commercial bank run out of Dallas. The core is business lending, commercial loans and commercial real estate, funded by business deposits gathered across a fast-growing state. Bolted on is mortgage finance, a $6.4B book that fronts cash to mortgage originators for the few days between when a home loan is made and when it is sold, low-risk and rate-sensitive. Since 2021, CEO Rob Holmes has torn up the old model and spent heavily building fee businesses: a broker-dealer doing investment banking and advisory, plus wealth management and treasury services. The moat, such as it is, is the Texas deposit franchise: cost of deposits fell to 2.40% last quarter versus 2.65% a year earlier. Tangible common equity is 9.9% of assets. The wager is that the fee build turns a middling bank into a better one before anything in the loan book breaks.

The numbers

The last five quarters show a bank past its trough but not compounding cleanly.

Q3 2025 was the high-water mark. Q2 2026's $1.83 was the first miss, four cents under the $1.87 estimate, after three straight beats, and net interest margin slipped to 3.28% from 3.35% a year earlier. The reason matters: earning-asset yields fell to 5.54% from 5.80% while deposit costs fell only to 2.40% from 2.65%, so asset yields dropped faster than funding costs. Deposits up 11% year-over-year papered over it.

The 2024 collapse to $1.28 was self-inflicted: a securities repositioning that produced a $61M net loss in Q3 2024 when Holmes repositioned the bond book. Painful medicine, and the kind of one-time hit that flatters every growth rate measured off it. 2025 snapped back to a record $6.79; 1H 2026's $3.39 tracks toward the $7.35 the Street models for the full year, which would be roughly flat.

On the money that matters for a bank, tangible book runs $77 a share and the bank earns about 10% on it (reported return on equity was 9.56% last quarter). At a 10% return, buying back stock at 1.3x book, $186M in 2025 and $75M more in Q1 2026, grows book value per share at a high-single-digit clip. That is the whole compounding engine, and it is adequate, not exceptional. What the market has not yet decided is whether the fee businesses lift that 10% toward peer returns; the single print that settles it is noninterest income, $75M last quarter and up 39% from a year ago, holding that pace.

Here is the tell. Noncurrent loans have nearly doubled in two years while the reserve against them has fallen from 1.23% to 1.07%. Charge-offs are still low in absolute terms, but the direction is wrong on both lines at once, and the non-accruals sit in commercial "business assets," $84M of them at June 30, the discretionary lending, not the low-risk mortgage warehouse. The filing shows the trend continuing through Q2 2026 with reserves at 1.08% while non-accruals remain elevated. Reserves keeping pace is the thing to watch.

Management

Holmes owns this rebuild, for better and worse. The uncomfortable data point: he sold $10M of stock across two sales in early June 2026, plan status not disclosed, so read it as discretionary until a Form 4 footnote says otherwise. Directors were small net buyers, Stallings putting up $663K in March. Capital allocation has been sensible, buying back stock at 1.3x book rather than reaching for deals, and pay is tied to the turnaround metrics he is chasing. Goodwill is effectively zero at $1.5M, indicating no M&A missteps. A large CEO sale near the finish of a five-year remake is not damning, but it is the one insider action that argues against the "nearly done, now it compounds" story.

How it fails or surprises you

Credit outruns the cushion. Noncurrent loans at 0.73% (as of Q1 2026) are rising, reserves at 1.07% are falling, and the bad loans cluster in commercial credit, not the safe warehouse book. A Texas CRE or business-loan crack forces a reserve build, and provisions eat straight into the 10% return. The print: next quarter's non-accruals against the reserve ratio.

The investment bank scales (right tail). The market pays almost nothing today for the broker-dealer Holmes built. If investment banking and wealth fees keep compounding near 39% and lift returns toward the 1.6x-book peers, the stock re-rates on both earnings and multiple. The print: noninterest income holding above a $75M quarterly run-rate through 2026.

Margin, not credit, is the near-term risk. The read least explained by my own numbers is Q2's miss: NIM fell even as deposit costs dropped, because asset yields fell faster. If rate cuts keep pulling loan yields down ahead of funding, net interest income stalls despite 11% deposit growth, and "steady compounding" quietly becomes flat. The print: NIM and earning-asset yield next quarter.

Closing thoughts

This is largely a priced stock, not a hidden one. At 1.3x tangible book for a bank earning about 10% on that book, the discount to the 1.6x peers is earned by the lower return, not an oversight; the payoff turns on whether the fee build lifts returns before credit bites. The fatter tail near-term is modestly to the downside, soft margin and creeping non-accruals in a flat earnings year, with the genuine upside sitting further out and depending on the investment bank actually scaling. That is a judgment, not a probability.

The bet is still that Texas Capital finishes its turnaround into steady profits: deposits growing, fees compounding, loans getting paid back. What breaks it is bad loans rising faster than the reserve set aside for them, and the one pair that tells you first is noncurrent loans against reserves-to-loans. If that gap keeps widening for two more quarters, the compounding story is wrong regardless of what deposits do.