RadNet, Inc. (RDNT) stock intrinsic value fair value estimate & owner earnings analysis

RadNet operates outpatient diagnostic imaging centers and, through its DeepHealth unit, develops artificial-intelligence and software tools for radiology. At the end of 2025 it operated 418 centers, directly or through hospital and health-system joint ventures, in Arizona, California, Delaware, Florida, Maryland, Virginia, New Jersey, New York and Texas; 368 of them are consolidated in its accounts and 101 of those are only partly owned. Medicare paid for 23.4% and Medicaid for 2.5% of imaging-center service revenue in 2025. The Digital Health unit had revenue of $92.7 million in 2025 (up 41%) and a loss of $30.3 million. The company has never paid a cash dividend. Revenue grew from $885 million in 2016 to $2.04 billion in 2025 and rose about 24% in the first half of 2026, to $1.20 billion. Cash left for owners has not followed. After interest, after setting aside swings in working capital, after stock compensation, and after the part of equipment and center spending needed just to keep the business running, it ranged from $0.09 to $0.88 a share across 2016 to 2025 with no upward trend ($0.38, $0.64, $0.46, $0.88, $0.35, $0.88, $0.28, $0.39, $0.56 and $0.09, in order). In 2025 operations produced $298.8 million of cash, of which $42.2 million was a working-capital inflow that we exclude. Stock compensation was $54.6 million, up from $29.8 million in 2024; we did not find an explanation in the filing, and first-half 2026 was $41.9 million against $37.2 million a year earlier, so we treat the higher level as lasting. Spending on equipment and centers was $213.3 million; we treat about $18 million of it as discretionary growth spending (using the ten-year average ratio of capital spending to revenue, 8.65%, applied to the year's revenue increase) and $195 million as the cost of staying in business. That leaves $6.9 million, or $0.09 a share. Depreciation and amortization are non-cash and are added back. Because the company has been buying AI businesses (for example iCAD in July 2025, an all-stock deal valued at about $110.7 million) while the Digital Health unit lost $30.3 million, we credit no growth from them. To value the shares we separate the business from its financing. We add back the $72.3 million of cash interest paid in 2025. We remove $31.9 million of interest earned on the company's cash, because the cash is counted separately below. We deduct $35.7 million of profit attributable to outside partners who co-own 101 centers; only $4.6 million was actually distributed to them in 2025, but the remainder, though not paid out, is theirs. On that basis 2025 comes to $11.6 million. Restating 2021 to 2025 the same way, with today's level of stock compensation and partner share, gives $10.8 million, minus $11.1 million, $20.0 million, $34.3 million and $11.6 million. We use the middle year, $11.6 million. 2020 is excluded (revenue fell 7% and working capital swung unusually), and years before 2020 describe a business with 57% or less of today's revenue. This measure fell by about two-thirds in 2025 (from $34.3 million to $11.6 million on today's cost basis). The base case assumes no change from the 2025 level; we found no dated event that reverses the decline, so we assume no rebound. What the company owes: at June 30, 2026 borrowings were $1.33 billion, up from $1.09 billion at the end of 2025 after about $249 million of new borrowing, in a half-year when the cash-flow data show roughly $316 million spent on acquisitions. Acquisitions the 2025 annual report lists as completed after year-end include RRC Royal Management (13 centers in Southwest Florida, about $65.0 million, January 2026), Northwest Radiology Network (6 centers in Indianapolis, about $9.0 million) and Gleamer SAS (completed March 2026). Cash was $726 million; we assume $100 million is needed to operate and treat the rest as available against the debt, giving net debt of about $706 million, or $8.97 a share. The revolver matures in April 2029 and the term loan in April 2031. A separate Truist credit facility at the New Jersey Imaging joint venture is one the company says it is neither borrower nor guarantor of; we use the consolidated borrowing figure without adjusting for it. We could not establish what the 2026 acquisitions earn, so no earnings from them are credited beyond what already appears in the results, while their cost is in the debt. Base case: the cash above stays flat at $11.6 million a year indefinitely. The business is worth about $193 million ($2.46 a share), against net debt of $706 million ($8.97), a difference of minus $513 million, or minus $6.51 a share. No growth is credited because ten years of strong revenue growth have not reached shareholders: equipment spending, stock compensation and partners' share appear to have absorbed it. Left out are earnings from the new acquisitions, a Digital Health turnaround and any margin gains, because none is yet shown in the cash flows. Shares cannot be worth less than zero; the negative figures show how far the debt exceeds what the evidenced cash flows support. Bear case: a repeat of 2020. Revenue fell 7.1% that year and the same measure of business cash (before interest, after partners' share) fell from $81.6 million to $44.2 million, a drop equal to 3.2% of revenue. On a revenue pace of about $2.4 billion that is a $78 million hit in year one and half that in year two, with recovery to $11.6 million afterward, as it did after 2020. The business is worth about $86 million ($1.09 a share) against $706 million of net debt, a difference of minus $621 million, or minus $7.88 a share. The company appears to owe far more than the business is worth here. The $626 million of available cash covers the two weak years, so we do not assume a failed refinancing. Medicare reimbursement cuts are an additional risk we do not size. Bull case: cash starts at $34.3 million (2024, the best of the five restated years) and grows 5% a year for ten years, about half the 9.7% annual revenue growth the company has sustained since 2016, including through 2020, then stays flat. The business is worth about $805 million ($10.23 a share) against $706 million of net debt, a difference of plus $99 million, or $1.26 a share. Testing the case: the largest uncertainty is how much equipment spending is routine. Depreciation of property and equipment was $136.1 million in 2025. If routine replacement equalled that rather than $195 million, cash available would be about $59 million a year higher and the base case would be about $6 a share. We kept the stated method because capital spending ran about 1.4 times depreciation and amortization in 2021, 2023, 2024 and 2025 (about 1.0 times in 2022), we found no split of routine and expansion spending in the filing, and the company cites replacing older units with newer AI-enabled equipment as a reason for rising equipment disposals. In 2016 to 2019 it ran at 0.9 to 1.0 times. If instead only the $4.6 million actually distributed to partners were deducted, the base case would be roughly zero a share. Only the equipment-spending case would make the shares worth something on the earnings shown to date, and it rests on a split the filings do not give us. The margin-of-safety price: we are not sure of the earnings level; the five restated years range from minus $11 million to plus $34 million, and we cannot tell whether 2025 is a low point or the new normal. The worst plausible outcome is a 2020-style shutdown or reimbursement cut while $1.33 billion of debt is outstanding; the structure survives it only because of the cash on hand, which is not locked away: the company has been spending on acquisitions. The shares pay no dividend, so any return must come from the value of the business growing. On today's evidence no price is supportable, because the base case is below zero. The margin-of-safety price therefore assumes a condition that would change this is true: net cash from operating activities, excluding working-capital changes, less capital expenditures and less stock compensation, of at least $100 million over four consecutive quarters (it was minus $11 million in 2025). At that level the business would produce roughly $105 to $135 million a year for shareholders and lenders after partners' share, which, held flat, supports about $16 a share. For a 20% annual return over an assumed five-year holding period from a business that pays nothing out and is still exposed to a 2020-type year, the price has to be near $8 to $9; we use $8 because one good year at that level would not prove it lasts. Our previous estimate was made on a different basis and its figures are not comparable. In business terms, this analysis reflects the June 30, 2026 balance sheet, with borrowings of $1.33 billion rather than $1.09 billion after the first-half acquisitions.