TERADATA CORP /DE/ (TDC) stock intrinsic value fair value estimate & owner earnings analysis

Teradata sells software for storing and analysing very large volumes of corporate data, used mainly by big banks, retailers, telecom carriers and airlines. Customers once bought the software outright bundled with Teradata hardware; today almost all revenue is recurring. At the end of 2025 annual recurring revenue was about $1.52 billion: roughly $701 million from Teradata's own public cloud service (up 15% in the year), $735 million from on-premises and private-cloud subscriptions (shrinking as those customers move to the cloud) and $86 million of old maintenance contracts (running off). Reported revenue has fallen from $2.32 billion in 2016 to $1.66 billion in 2025 as consulting and one-time sales were run down. In the first half of 2026 revenue was $854 million against $826 million a year earlier, so the top line has stopped falling for now. How we measured earnings. We start from operating cash flow and remove three things: working-capital swings, which are large because annual renewals are invoiced around year end (2022 was flattered by a $50 million tax refund); stock compensation, which ran $112 to $126 million a year in 2021-2025 and was $62 million in the first half of 2026, a real cost to owners even though no cash leaves; and capital spending including repayment of finance leases ($69 to $92 million a year since 2020), because those leases pay for the servers Teradata installs to deliver its service. Adding back cash interest so the business is valued before its lenders, it generated about $89 million in 2020, $218 million in 2021, $62 million in 2022, $110 million in 2023, $139 million in 2024 and $189 million in 2025. We measure from 2020, after the switch from perpetual licences to subscriptions; earlier years describe a different business. The median of those six years is about $125 million, roughly $1.30 a share, and that is our starting level. It sits well below 2025, which grew about 36% on lower stock compensation, favourable currency and working-capital timing; we do not assume that pace continues and start the base case flat. The years are lumpy, so our confidence in the level is moderate. The balance sheet. In February 2026 Teradata settled its trade-secret lawsuit against SAP for $480 million gross, about $355 to $362 million before tax after legal fees, and used the cash to repay its entire $456 million term loan. At 30 June 2026 it had $414 million of cash and no bank or bond debt. We treat about $150 million as needed to run the business, leaving roughly $264 million, about $2.75 a share, as surplus. Finance leases of about $91 million are not subtracted, because their repayment is already charged in the earnings figure. The settlement is counted once, as cash, never as earnings; the large first-half 2026 profit and operating cash flow are mostly that settlement. Base case. Recurring revenue roughly holds as cloud growth offsets the shrinking on-premises base, as it did in 2024, 2025 and the first half of 2026, so business cash flow stays near $125 million for three years. From 2029 it declines about 3% a year and keeps fading after year ten, continuing the long-run revenue slide, with no growth credited: cloud expansion within existing customers is real but has so far only replaced migrating revenue. The business is worth about $1.47 billion; with no debt and $264 million of surplus cash, about $18 a share. Bear case. A repeat of the 2019 and 2022 troughs, when the business generated only about $61 to $62 million, triggered by an enterprise IT spending downturn like 2009 (Teradata's revenue fell roughly 8% that year) or a wave of on-premises customers leaving for Snowflake, Databricks or the cloud providers' own warehouses rather than moving to Teradata's cloud. Two years at about $61 million, a partial recovery to about $92 million, then a 7% annual decline that continues indefinitely as the base shrinks while stock compensation stays high. The business is worth about $740 million; with no debt and the surplus cash, about $10.45 a share, of which about a quarter is cash. Bull case. The 2025 level of about $175 million holds for ten years because cloud revenue keeps replacing the on-premises decline and the maintenance tail finishes running off, then fades about 3% a year. No acceleration is credited. The business is worth about $2.34 billion, about $27 a share. No case carries a debt burden, so none depends on refinancing; the $400 million revolving credit line is undrawn. Compared with our previous estimate the base and margin-of-safety price are unchanged, the bear case is slightly higher and the bull case lower: we no longer credit the bull case with a permanent 2025-level business beyond year ten, because nothing we can point to shows the on-premises decline ending. Margin-of-safety price. We are only moderately sure of the earnings level: six years ranged from $62 million to $218 million, and stock compensation alone consumes about as much as owners keep. The worst plausible outcome is that Teradata's largest long-standing customers consolidate onto a hyperscaler's native warehouse, turning a slow decline into a fast one; the cloud net expansion rate already slipped from 117% in 2024 to 108% in 2025. The structure survives that: no debt, real cash, and positive cash generation even in the worst years since 2019. If we are wrong about both the level (closer to $90 million) and the decline (closer to 7% a year), the shares are worth about $11 to $12. There is no dividend; owners are paid through buybacks, which cut the share count from 131 million in 2016 to about 93 million today and absorb essentially all free cash. The price assumes a three-to-four-year holding period. At $9 a share the price values the operating business at about $6.25 after surplus cash, about five times its median annual cash generation, so the bear case is worth more than that price and the base case implies well over 20% a year within three years. The margin-of-safety price is no higher, because the base case itself describes a business in slow decline and a higher price would make the 20% target depend on a stabilisation not yet demonstrated.