Teradata Corporation TDC

Key data

The business

Teradata sells the data warehouse that huge organizations use to store everything they know and ask hard questions of it, the airline pricing a seat, the bank scoring a loan, all against decades of history in one place. For most of its life that database ran on hardware in the customer's own building, sold as a perpetual license. The whole company today is a forced march to move those same customers onto VantageCloud, a subscription that runs on Amazon, Microsoft or Google instead. The moat is switching cost: a Fortune 500 company does not re-plumb its system of record on a whim, which is why revenue erodes slowly rather than collapsing. Recurring revenue is now 89% of the total, the perpetual license is nearly extinct, and low-margin consulting is being deliberately shed. What you own is a shrinking top line with a rising quality of earnings underneath it.

The numbers

The last five quarters show the tension: a flat business with one quarter that lies about it.

That $335 million first quarter came almost entirely from a $472 million non-operating gain; operating income that quarter was actually negative $36 million. The gain the last look couldn't trace to a source appears in the 10-Q as other income, and it held as suspected: strip it out and Q1 was an operating loss, not a windfall. Normal quarterly earnings sit near $0.45.

Revenue has fallen about 4% a year for five years. The interesting line is underneath it. Here is Q2 by type:

Across the first half, recurring revenue grew from $712 million to $763 million, up 7.2%, while gross margin rose to 59.3% from 56.4% on that richer mix. The whole decline in the top line is consulting the company chose to walk away from. The compounding math is the legacy-software playbook: revenue melts, but per-share value holds because margins climb and the share count falls. Buybacks of roughly $1.3 billion over 2021 to 2025 took shares outstanding from about 108 million to 94 million. Normalized free cash flow ran near $286 million in 2025, about $3.04 a share against a $28 price, an 11% yield on cash that does not need the one-timer to exist. The variant the market is not paying for: recurring revenue may have already stopped shrinking. The print that settles it is cloud ARR growth over the next two quarters.

Management

CEO Steve McMillan sold $1.5 million of stock in February, CFO Richard Petley $1.06 million in May, and holder Lynrock Lake $1.6 million; 13 insider sales, zero buys, $8.3 million total over twelve months. Form 4 plan status is not disclosed, so planned and discretionary sales cannot be split, but the dollar amounts are routine for officers of a company this size. The louder signal is capital allocation. In the first half management repaid all $431 million of long-term debt and still bought back $74 million of stock, leaving the balance sheet debt-free with $414 million cash. Non-GAAP earnings have topped consensus four quarters running, which buys the guidance some credibility. Stock compensation, near 7% of revenue, is a real cost that eats roughly $120 million of the cash the buyback returns.

How it fails or surprises you

On-prem erosion wins. If recurring growth stalls back below 3% and total revenue resumes its 4% annual slide into 2027, the melting-ice-cube price is correct and there is no re-rate. The tell is recurring revenue growth; two soft quarters end the story.

Recurring is genuinely inflecting (right tail). First-half recurring grew 7.2% while gross margin crossed 59%. If cloud ARR keeps compounding double-digit and mix lifts margin past 60%, free cash flow per share re-rates off 10x forward earnings, and today the market pays nothing for that. The print is cloud ARR plus gross margin.

The cash is not as big as it looks. The $472 million gain and a $401 million first-quarter operating cash haul make trailing free cash flow read near $745 million. If that cash was one-time, normalized free cash flow is about $286 million, and the "28% yield" case collapses to 11%. Full-year 2026 operating cash flow, ex the gain, proves it either way.

Closing thoughts

Recurring revenue growth over the next two quarters settles it. Above 5%, the business has stopped shrinking and you are getting a debt-free cash generator at 10x forward earnings buying back its shares. Below 3%, this is a slow liquidation priced correctly. An ambiguous middle, flat recurring with margins still climbing, favors the owner because the buyback works while you wait. The downside is a value trap eroding a few percent a year; the upside is a re-rate on earnings the market has written off. With 7% recurring growth confirmed against a decade-low multiple, the fatter tail points up, but the one-timer flatters the cash and demands normalizing first.

The bet is still that the banks, airlines and telcos keep moving Teradata's database to the cloud rather than ripping it out, and that the growing subscription revenue outruns the shrinking old software. It breaks if recurring revenue growth rolls back under 3% while gross margin stops climbing; those two numbers, watched together, tell you first.