First American Moves the AI Payoff to 2028 — Right as the Buyback Goes Quiet

The date moved

For five straight quarters, First American's story had a reliable shape: endure the residential trough, roll out Endpoint and Sequoia, decommission the legacy stack, and let the title margin climb. The shape survived this call. The date did not.

Asked directly whether 2027 holds margin expansion, CFO Matt Wajner was blunt: “I would say the margin right now, I don't see a lot of room for expansion.” — Matt Wajner, Chief Financial Officer · 2026-10-22 That is a different sentence than the one he offered three months ago, when he told Oscar Nieves, “when we look at the back half of the year, I think we can expand on that.” — Matthew Feivish Wajner, Chief Financial Officer · 2026-07-23 Same company, same platforms, one quarter apart.

The full-year framing is equally stark — essentially flat revenue versus 2026, commercial up modestly to a record, residential down, investment income up, and margins stable-to-slightly-contracting. Mark Seaton then dated the payoff explicitly: “we will start to really kind of start to see the benefits in 2028 if we can successfully execute next year.” — Mark Seaton, Chief Executive Officer · 2026-10-22

Our outlook today is essentially flat for revenue year-over-year when we look at 2027 compared to full-year 2026... Obviously, the residential market, we're more pessimistic about both for purchase and refi.

This matters because First American spent a year telling investors that the roughly 100 basis points of technology drag was an investment, not a cost. In October 2025, Seaton framed it as arithmetic: “you don't spend 100 basis points just to get nothing.” — Mark Seaton, Chief Executive Officer · 2025-10-23 That bet is still live — but the return is now two years out, not one.

What actually improved is measurable

Credit where due: the operating metrics are not vapor. Endpoint's automation rate went 30% in Q1, 34% in Q2, 42% in Q3. Average human review time on Sequoia purchase files fell from 83 minutes to 74 to 56. Sequoia now completes 17% of California purchase transactions processed through Endpoint, up from 11%, and 34% of California refis. The stated target is roughly 80% of refis and 70% of purchases in title-plant markets.

But the expense side is still the old problem. As Seaton conceded on the October 2025 call, “we're really double paying with technology right now.” — Mark Seaton, Chief Executive Officer · 2025-10-23 And the risk register has genuinely rotated. The company's own top keyword this quarter is big risk — and the second is change management, which is new. Years of technological uncertainty have been replaced by the very human problem of retraining escrow officers who have used the FAST system for a quarter century. The bottleneck moved from code to people. That is progress, but it is not the kind of progress that shows up in a margin line quickly.

The reserve lever nobody priced

The quietest number on the call may be the most important. The provision was 3.0% of title premiums, composed of 3.75% for the current policy year and a $11 million net release from prior years. First American has run below its 4%–5% cycle loss rate for several years, and Wajner all but flagged the turn: “We would think of it in the current environment for new policies as 3.75% is kind of the right loss reserve right now.” — Matt Wajner, Chief Financial Officer · 2026-10-22 He said something similar in February, but with less conviction about timing. The word reserve is a new entrant to the company's top keyword set, and it deserves to be. If the 75 basis points of release stops, the reported loss ratio rises 75bp against a title margin of 12.4% — roughly the entire difference between "stable" and "contracting."

That is the honest read on an operating margin of 13.4%, up 2.8 points year over year but still 34% below the 2021Q3 peak of 18.5%. The recovery is real; the next leg depends on a reserve policy, not a software release.

Commercial carries the quarter, and data centers shrink as a story

Commercial was genuinely strong: $297 million of revenue, up 21%, with average revenue per order at a record for a fifth straight quarter and eight of eleven asset classes growing. Seaton closed 22 deals over $1 million of premium versus 12 a year ago — but only six were data centers, and he was explicit that energy, industrial and multifamily carried the rest. Fittingly, data center is one of the largest decliners in the company's own keyword momentum this quarter, down sharply after five quarters as the headline growth theme. That is a de-risking of the narrative, not a deterioration of the business — a broadening that reduces concentration but also removes the most exciting slide from the deck.

The durable point is duration. “it's just not as sensitive to interest rates... A lot of these commercial loans are shorter in duration now, too.” — Mark Seaton, Chief Executive Officer · 2026-10-22 Shorter maturities mean more frequent refinance events, which is the structural offset to a frozen purchase market.

Capital: a raise and a retrenchment

The dividend went up 11% to $2.44 annually, moving toward the 40% payout ratio target. The buyback went the other way. First American repurchased 23,000 shares for $1 million in the quarter and 143,000 shares for $9 million in October at an average of $62.40 — against a stock that is 17.8% below its July peak of $79.16. On the July call, Seaton had already said “we are not in the market at the moment.” — Mark Edward Seaton, Chief Executive Officer · 2026-07-23 That was a pause. This is now a pattern, executed at prices well below where the company was buying earlier in the cycle.

Repurchases of $20 million against $219 million of net income is a buyback payout of about 9%, versus a program that once ran to $227 million in a single quarter. The justification given in February — building "dry powder" for AI-era optionality — still stands, but shareholders are being asked to accept a smaller return of capital while waiting for a 2028 margin.

The macro wave is not FAF's friend on volume

There is a global confluence here worth naming. The market's own top themes this quarter include Fed rate hike, mortgage universe and retail deposits — the same rate regime that has pushed the 30-year fixed back to 7.5%, its highest since mid-2024, and that shows up in First American's October order book as purchase down 8%, refinance down 35%, and commercial down 7%. The company is on the wrong side of that wave for transaction volume, which is why “longer residential origination environment” is now a standing phrase rather than a scenario.

It is on the right side for the float. Title investment income grew 11%, and net of interest expense it grew 8%, with Wajner guiding that 8% into Q4 and further growth in investment income in 2027 if rates stay higher. The catch is that this offset is financed with a growing deposit book, which is why interest coverage of 6.2x sits so far below the 31.4x of 2021.

The bear case writes itself: flat revenue, no margin help, an earnings base still flattered by reserve releases, and a buyback on hold. The bull case is simpler, and it is the one management is asking for time on — 60 basis points of market share gained over twelve months with zero acquisitions, a commercial franchise that does not care about the 30-year mortgage, and two platforms that finally have measurable automation numbers. The market has already marked the stock down 17.8% from its summer high while it waits to see which one is right.