GAXOS.AI INC. (GXAI) stock intrinsic value fair value estimate & owner earnings analysis

Gaxos.ai is a very small technology company (three full-time employees as of March 2026) that has changed direction several times since it listed. It began as an NFT and game developer and now sells AI subscription apps (Art-Gen.ai for images and video, UnGPT.ai for rewriting AI text, and Bible Pray AI) and, through a 70%-owned subsidiary called RNK Health, provides administrative support for a wellness and peptide business. In 2025 revenue was $1.93 million, of which $1.51 million came from RNK Health's administrative services and $0.42 million from the app subscriptions. Revenue in the first half of 2026 was $4.27 million, up sharply from about $0.19 million a year earlier. The company has never produced positive operating cash flow: it used $1.3 million in 2022, $3.0 million in 2023, $3.3 million in 2024, $3.9 million in 2025 and $3.6 million in the first half of 2026. The operating loss for the first six months of 2026 was $4.4 million. Its reported net loss for that period was smaller only because of a roughly $1.75 million non-cash gain on the exchange of investments, which we do not treat as earning power. How we measured earnings. We measured what the owners actually earn: operating cash flow, less the swing from working capital, less stock compensation, less only the capital spending needed to stand still. On that basis owner earnings were about negative $1.63 per share in 2022, negative $3.95 in 2023, negative $1.97 in 2024 and negative $0.58 in 2025 (the 2025 figure looks better per share only because the share count rose from about 1.8 million to about 7.1 million as the company sold stock; in total dollars the loss grew from $3.5 million to $4.1 million). We used 2024 and 2025 as the relevant window because that is when the company became an AI and health business; earlier years describe an NFT-gaming startup. The company has no borrowings, so there is no interest to add back and no debt to subtract. The trailing year got worse in total dollars and we credit no improvement in the first year of any case. Capitalised software and the 2025 software purchase were treated as growth spending because revenue jumped from almost nothing, so no maintenance spending was deducted. What the shares are worth. Nothing in the filings gives us a named, evidenced source of future positive earnings. The revenue ramp is real (second-quarter 2026 revenue of $2.46 million against $1.81 million in the first quarter), but the operating loss grew alongside it, the health revenue is service fees from a partnership that is 30% owned by an outside party, and the apps are new. Under our approach, growth that depends on new products and markets that have not yet proven themselves is left out, so the operating business itself is worth nothing. What the shares are worth is therefore what is left of the company's cash after it finishes spending it. At June 30, 2026 the company held about $11.9 million of current assets (cash of $1.09 million plus short-term investments and receivables) against total liabilities of $1.13 million, none of it borrowed money, for net liquid assets of about $10.8 million, or roughly $1.00 per share on 10.7 million shares outstanding. That cash was largely raised by selling about 3.1 million new shares for $5.4 million in early 2026. The company also paid $2.9 million in March 2026 for a 19.99% stake in America First Defense.AI LLC, a private defense-technology company; nothing in the filings allows it to be valued, so it is worth zero here and is upside if it proves out. Bear case, about $0.11 per share. The business keeps burning cash at its first-half 2026 pace of about $7.9 million a year for roughly 15 months before management is forced to stop. The business is worth nothing, it owes $1.1 million of ordinary bills (no debt), and the burn consumes most of the roughly $10.8 million of liquid assets, leaving about $1.2 million or $0.11 per share. Base case, about $0.20 per share. Losses run at roughly $6 million a year, between the 2025 loss of $4.1 million and the current $7.9 million pace, for about 18 months, after which spending is cut back to what the remaining cash can support. The business is worth nothing, it owes $1.1 million of ordinary bills, and after roughly $8.6 million of present-value burn about $2.2 million remains, or $0.20 per share. Bull case, about $0.49 per share. The growth in health-services and app revenue seen in the last two quarters keeps narrowing the loss so that burn falls to about $3 million a year (below 2025's $4.1 million) for two years before the company reaches a self-funding size. That is $5.5 million of present-value burn against $10.8 million of liquid assets, leaving about $5.3 million or $0.49 per share. We give no credit for profits beyond that, because none is evidenced. What could make this wrong. Management says its cash will cover the next twelve months, and there is no going-concern warning, but the company also states that it will need additional financing and that its revenue does not cover its costs. If it raises money again it will most likely sell shares at low prices, which would reduce the value per existing share below our figures. On the other side, the America First Defense stake and the health and app businesses could be worth far more than nothing, and if revenue keeps compounding the way it did in the first half of 2026, the loss could disappear faster than the bull case assumes. The company also carries a $1.00 Nasdaq minimum-bid requirement; no deficiency notice appears in its latest annual report. Warrants held by investors at $2.33 to $3.32 per share would add cash and shares only if the stock rises well above our estimates, and we have not counted them. We could not reconcile the size of the first-half 2026 operating cash outflow with the smaller reported net loss from the filings alone; the gap is explained by the non-cash investment gain, but details of the investment portfolio in the latest quarter were not available.