Rank One Computing Corp (ROC) stock intrinsic value fair value estimate & owner earnings analysis

Rank One Computing (ROC) is a Denver-based developer of face and fingerprint recognition software whose algorithms are regularly ranked by the National Institute of Standards and Technology among the most accurate and efficient in the world. It licenses the ROC SDK to companies that build it into their own systems ($5.8M of 2025 revenue), sells ROC Watch video analytics ($5.8M in 2025, up from $1.3M), the ROC ABIS large-scale biometric search system, ROC Enroll and the newer ROC Evidence, and earns money from government-funded research contracts ($4.9M in 2025). 2025 revenue was $17.0M, up 24%. Customers are mostly US federal, defense and law-enforcement agencies, and two customers made up 43% of 2025 revenue. The company was an S-corporation until 2024 and listed on Nasdaq in February 2026, raising about $21.5M net at $6.00 a share. Earnings record. On a cash basis the business has not yet made money for its owners in any year on record. We measure owner earnings as operating cash flow with swings in receivables, payables and deferred revenue removed, minus stock compensation and the capital spending needed to stay in business (including capitalized software). On that basis owners lost about $1.36M in 2024 (-$0.09 a share) and $2.12M in 2025 (-$0.14 a share); both reconcile to net income plus non-cash charges. In the first half of 2026 operating cash flow was -$6.0M, of which about $2.9M was working capital (customers paying more slowly, payables and deferred revenue paid down). Stripping that out and deducting $0.47M of stock compensation, $0.86M of equipment (largely data-center build-out) and $1.09M of capitalized software leaves an owner loss of roughly $5.6M for the half. The two quarters were very different. The first quarter was poor: revenue of $2.5M, down 20%, as government research funding stalled, and an operating loss of $3.0M. The second quarter recovered: revenue of $5.1M, up 2%, with research-contract revenue up 41% on a $4.9M follow-on award, a 90% gross margin, and an operating loss of only $0.8M. Product revenue fell 26% in that quarter, however, so the recovery came from contract timing rather than the software lines. In August the company signed a five-year federal ABIS contract worth $5.1M including options, of which $2.5M is committed. With only two audited years and no year of positive owner earnings, there is no level to normalize to, and we do not credit a profit the business has never reached. What changed since our previous estimate. The second quarter showed the first-quarter loss rate was not the run rate: operating expenses held near $5.3M a quarter while revenue doubled sequentially. We have therefore assumed a somewhat smaller loss in the coming year (about $5.5M rather than the first-half pace of roughly $11M a year, which included a non-recurring dip in research revenue and data-center spending). That raises the base case slightly. We now assume the bull case's earning power fades after year ten rather than lasting indefinitely, because a ranking lead in algorithms is not, on the evidence, a durable moat against much larger competitors; that lowers the bull case. Balance sheet. The 9% credit line ($1.84M at the end of 2025) was repaid in full. At June 30, 2026 there were no borrowings and $11.9M of cash, about $0.62 a share on 19.08M shares. Interest paid was trivial, so valuing the business before and after lenders gives the same answer. About 9.4M stock options with an average exercise price near $2.12 are outstanding; they are worthless at any of our case values and are not counted as dilution, but they would take a large share of any gain above that price. Before the IPO, management had concluded there was substantial doubt about the company's ability to continue as a going concern (it held $72K of cash at September 30, 2025); the IPO proceeds resolved that, but at the first-half 2026 rate of cash use the cash covers well under two years. Bear case ($0.00). Federal purchasing is the named cyclical risk, and this company has already lived through one squeeze, when 2025 cash fell to near zero and the first quarter of 2026 lost a fifth of its revenue to a funding slowdown. The federal record includes the 2013 sequestration and repeated shutdowns and continuing resolutions that froze new awards for quarters at a time. In this case research revenue relapses, one of the two large customers (25% and 18% of revenue) does not renew, and losses run about $8.5M, $6M and $3M over three years. Those losses are worth about -$15.9M today against $11.9M of cash; the business owes nothing, but the cash runs out in 2027 and a raise at depressed prices or a weak sale leaves today's shares with essentially nothing. Business value: below the cash it consumes. Debt: none. Equity: about zero. Base case ($0.51). The revenue base holds near $17M with the new ABIS contract and the second-quarter research award, and losses shrink as the one-time costs of listing and the data-center build pass: about $5.5M in year one, $2.5M in year two and $0.5M in year three. Year one is still deeper than the 2025 loss of $2.1M, consistent with actual 2026 results. From year four the business earns about $1M a year, close to what its 2024 cost structure implied, through year ten, then fades to zero by year fourteen because nothing we can name keeps the lead durable beyond that. The losses are worth about -$7.8M today and the later earnings about +$5.7M, so the business is worth about -$2.2M; adding $11.9M of cash and subtracting no debt gives about $9.7M, or $0.51 a share. Revenue growth is not credited: the 2025 jump came from one product's ramp, product revenue fell in the second quarter of 2026, and there is no backlog disclosure or multi-year compounding record. Growth in the 2025 to 2026 owner loss (from -$2.1M to a first-half annualized rate near -$11M) is worse than our year-one assumption; the reason for assuming improvement is dated and in the filings: the second-quarter 2026 operating loss of $0.8M against $3.0M in the first quarter. Bull case ($1.76). The company's own levers work: the ABIS award and research follow-ons renew, ROC Watch converts pilots to deployments, and expense discipline holds. Losses of $3.5M and $1M give way to $1M of earnings in year three and about $3M a year from year four (roughly 15% of a $20M revenue base at a gross margin near 85%), fading over the following twenty years. The business is worth about $21.7M plus $11.9M of cash, no debt, or about $1.76 a share. What would make this wrong. Results are lumpy and depend on the timing of federal awards; one large award could change the picture quickly, and the strong second quarter could prove the start of a turn. Equally, product revenue fell in that quarter and nothing in the filings yet shows the loss narrowing on a sustained basis. The biggest unknowns are whether the two large customers renew and whether research funding stays steady. A March 2026 lawsuit filed in the High Court of England and Wales by Eye Corp IT Solutions names the company; management does not expect it to be material. Margin of safety price. Our confidence in the earnings level is low because there isn't one: no year shows positive owner earnings. The worst plausible outcome is a cash run-out in 2027 followed by a heavily dilutive raise, and with this much customer concentration and dependence on federal budgets, nothing in the structure of the business would shield shareholders from it; the absence of debt only means no lender forces the issue. The business pays nothing during the holding period. A 20% annual return requires a business that has stopped consuming cash, so the margin-of-safety price applies only after two consecutive quarters at or above break-even on owner earnings without new equity. By then roughly $8M of cash would likely remain, and $0.75 a share would price the whole company near $14M. A break-even software business with about $17M of high-margin revenue and top-ranked algorithms, earning even $1M to $2M a year by 2028 to 2030, should be worth well over twice that, which is about what a 20% annual return over four to five years requires. Until that condition is met, no margin-of-safety price applies.