$92.00
$215.00
bear $92.00
base $158.00
bull $215.00
margin of safety $75.00
market price* $88.14
Axos Financial is a branchless US bank holding company based in San Diego, with about $30 billion of assets at June 30, 2026. It gathers deposits online and through specialty channels ($24.6 billion at year-end) and lends through national niche businesses. At June 30, 2026 the loan book was $26.1 billion: $9.5 billion of commercial and industrial loans (lender finance, fund finance and equipment leasing), $8.9 billion of commercial real estate, $4.6 billion of single-family mortgages and mortgage warehouse lines, $2.5 billion of multifamily and commercial mortgages, and $0.7 billion of auto and consumer loans. It also owns a securities clearing and custody business, Axos Clearing, and a small advisory arm. With no branches, it runs at a much lower cost than branch-based banks, and that cost advantage has let it compound equity at mid-to-high-teens rates for a decade. Diluted shares fell from 63.6 million in fiscal 2017 (June year-end) to 57.8 million in fiscal 2026. What changed since our last look. On September 21, 2026 Axos closed its purchase of IRA deposits from Capital One. The deposits transferred were about $1.9 billion, not the roughly $3.2 billion described when the deal was signed in April. Together with the $2.3 billion of Jenius Bank deposits taken over from SMBC in May 2026, the bank has added about $4.2 billion of funding this year. We do not give these deposits any value of their own: acquired deposits only earn money once the bank lends them out, and that lending is already inside the growth we credit below. Our earlier figures were built on a different basis and are not directly comparable. How we measure earnings. For a bank, operating cash flow adds back the provision for credit losses. We deduct actual net charge-offs instead, because loans that go bad are a real cost. We strip out swings in mortgage loans held for sale and in the clearing firm's customer balances, deduct stock compensation, and deduct maintenance spending on premises, technology and leased equipment. We exclude the $92 million gain Axos booked in fiscal 2024 on buying about $1.3 billion of loans from the FDIC at a discount, and we exclude the accretion of that purchase discount ($91 million to $108 million a year in fiscal 2025 and 2026), because it is a pool that runs off. The $125 million purchase of a San Diego office campus in January 2026, for a future headquarters, is treated as one-time. On this basis owner earnings were $5.98 per share in fiscal 2024, $5.31 in 2025 and $8.66 in 2026. Fiscal 2026 was lifted by $93 million of deferred taxes, a timing benefit that reverses; fiscal 2025 carried a $28 million deferred-tax charge. Without those, the two years were about $5.80 and $7.05 per share. The starting level. Because the bank's equity has grown about 14% a year, older per-share figures describe a smaller bank. We therefore measure normal earning power as a return on the capital the bank has today. Over fiscal 2024 to 2026, the years at roughly its current shape after the FDIC loan purchase, owner earnings (excluding the deferred-tax swings) were about 15% to 16% a year on tangible equity. Tangible equity is about $50.40 per diluted share at June 30, 2026 ($3.17 billion of equity less $249 million of goodwill and intangibles). We start the base case at a 15% return, slightly below that record, because credit costs have been unusually low: net charge-offs were 0.05% of loans in fiscal 2024, 0.14% in 2025 and 0.18% in 2026, rising each year and led by commercial and industrial loans. That gives about $7.57 per share in the first year, versus $7.05 for fiscal 2026 after removing the tax benefit (a 7% rise, against 22% growth on the same basis in fiscal 2026). The key judgement: growth costs capital. A bank cannot grow its loans without growing its equity alongside. Axos Bank's capital ratios fell during fiscal 2026 as loans grew 21% (common equity tier 1 from 12.42% to 10.93%, tier 1 leverage from 10.23% to 9.24%), so there is little spare capital. In every case, the equity needed to support growth is held back from owner earnings before anything counts as cash for owners. Axos pays no dividend; that cash comes back as buybacks or as a larger bank. Debt. For a bank, deposits and borrowings are the raw material of the business, and interest is an operating cost already inside owner earnings, so no debt is subtracted. The holding company's $343 million of subordinated notes is likewise already paid for in the earnings stream. Leverage is charged in the bear case below. Base case, about $158 per share. The return on tangible equity drifts from 15% to 13.5% over ten years, reflecting a net interest margin that fell from 4.90% to 4.70% in fiscal 2026, an efficiency ratio that rose from 43.6% to 49.5% since fiscal 2024, and charge-offs settling near 0.30% of loans. We credit equity growth of 8% a year for five years, tapering to 3% by year ten. That growth is the continuation of the bank's own long-running process. Axos has grown its loan book by well over 15% a year since 2009, kept growing through the 2008–2010 crisis, and funded it from retained earnings while shrinking its share count. We credit about half that historic pace and taper it as the bank gets larger. We do not credit faster growth from the newly acquired deposits, new lending niches or a wider margin. After year ten we assume no further growth. The bank keeps earning 13.5% on its capital and pays it all out. What sustains that is the same branchless deposit franchise and low cost base that have produced above-average returns for twenty years. Almost all of this value is the business itself; there is no debt to subtract beyond what is already counted. Bear case, about $92 per share. A full credit cycle sized from the industry's 2008–2011 record. At FDIC-insured banks, annualized charge-offs on real-estate loans ran about 2% to 2.7% through 2009 and 2010, and about 1.7% on income-producing commercial property, peaking in 2009–2010 before easing in 2011. Axos's real-estate loans were made at about 50% of appraised value on average, which should limit losses compared with the industry. Against that, 37% of its real-estate lending is in New York, including 53% of its commercial real estate (up from 37% a year earlier) and 39% of its multifamily loans. Its lender-finance and fund-finance books have grown fast and have never been through a recession at this size. We charge net losses of 1.0%, 1.75%, 1.5% and 0.75% of loans in years three through six: about 5% of the loan book cumulatively, or roughly $1.3 billion at today's size. We also raise the loss reserve from 1.34% to 2.5% of loans in the first year of the cycle, releasing a little afterwards. Before losses, the return on capital falls to 11% to 12% as rates and lending volumes drop, and growth stops for four years. In the two worst years, earnings before credit losses roughly cover the losses and the reserve build, so capital stays near today's level and no new shares are needed. After the cycle, the bank grows 3% a year and settles at a 12% return, then 11.5% once growth stops. This case is worth more than the $50 of tangible book because a 2008-sized cycle consumes about three years of earnings rather than the bank's capital. Its low cost base keeps earnings before credit losses near 2% of assets, enough to absorb losses on that scale. Bull case, about $215 per share. The return on tangible equity starts at 16.5% and eases to 15%. That is the level Axos reached in fiscal 2024 and on reported numbers in fiscal 2026. Equity grows 12% a year for five years and tapers to 5%, still below the bank's record, as its national niches keep growing at the pace they showed in fiscal 2026. After year ten it grows no further and pays out its earnings. What would make this wrong. Most likely, losses on the New York real-estate book or in lender finance and fund finance arrive sooner or deeper than the bear case allows. Profitability is also slipping: the margin is narrowing, costs are rising, and return on equity fell from 21.6% in fiscal 2024 to 16.3% in fiscal 2026. If returns settle at 12% and growth at 5%, the shares are worth about $119. If that happens and a 2008-style cycle also arrives, they are worth about $84. Online deposits can leave quickly in a panic, as US regional banks saw in March 2023. Axos came through that episode, but a run on its deposits is a risk none of the cases fully models. A shareholder suit dating from 2015 was dismissed in September 2025 and is under appeal. It seeks damages on the company's behalf, not against it. Not reconciled. Office lease payments could not be fully separated in every year of the cash-flow statement. This may overstate owner earnings by $10 million to $15 million a year, under 4% of the starting level. We did not credit the $153 million of unaccreted loan discounts and fees on the balance sheet. Margin of safety price: $75. How sure are we of the earnings level? Only moderately. Every year we can measure came in a benign credit period, charge-offs have risen three years running, and a fifth of the loan book was made in the last twelve months. The worst plausible outcome is a deep New York property downturn arriving at the same time as a deposit run. A bank levered about ten to one with a 10.9% common equity ratio would be forced to shrink and might have to issue shares near the bottom. The 50% average loan-to-value is what makes that survivable rather than ruinous. If we are wrong about returns and growth, the shares are worth about $119; if we are also wrong about the cycle, about $84. Axos pays no dividend, so any return over the holding period comes from the bank getting bigger and buying back shares. The price therefore assumes value is recognized within about three years. From $75, recognition of the base case within three years implies more than 30% a year. The lower-return path implies about 20% a year over the same time. Only the combination of lower returns and a full credit cycle falls short, and even then the business is still worth more than $75. Above about $75, the lower-return path stops clearing 20%, and that path is too likely to ignore for a bank lending this fast into concentrated markets.