The five largest U.S. banks just told the market something revealing: the office loan losses they spent two years preparing for are not arriving at the scale they feared. JPMorgan Chase, Goldman Sachs, Bank of America, Citigroup, and Wells Fargo all beat second-quarter earnings expectations and, more importantly, cut their provisions for credit losses. Goldman slashed its PCL by 38% quarter-over-quarter to $102 million. Bank of America and JPMorgan each cut by more than 10%, to $1.4 billion and $2.5 billion, respectively. Wells Fargo's allowance ticked up modestly, but CFO Mike Santomassimo explicitly credited improvement in the bank's commercial real estate office loan portfolio for offsetting growth in credit card and auto loan reserves.
The market meaning is not that the office distress cycle is finished. It is that the largest banks have already recognized the losses they expect to take on their pandemic-era office loans. The reserves they built over the past four quarters were sufficient. The repricing has happened on their books. The question now is whether the same can be said for the regional banks, the private credit lenders, and the CMBS trusts that still hold a large share of the office debt maturities coming due through 2027.
The banks' confidence is grounded in two things: a resilient transaction market and a massive new source of capital demand. May 2026 saw a bounce-back in commercial real estate sales volumes after an April collapse. That is not a recovery in office fundamentals. It is a recovery in liquidity. Sellers who waited for a bid are finally getting one, and buyers who demanded a basis reset are finding assets that pencil at current interest rates. The banks are not lending into that market at the same volume as 2021, but they are seeing their existing loans get paid off, restructured, or sold at prices that do not force additional loss recognition.
The second factor is the AI and data center buildout. Bank of America CEO Brian Moynihan dismissed the idea that legacy CRE problems would surface at the scale once feared. JPMorgan CFO Jeremy Barnum pointed to a global economic reordering that includes not just AI but infrastructure, remilitarization, and trade restructuring. The banks are not underwriting data center loans as a substitute for office loans. They are underwriting them as a new, large, and capital-intensive asset class that diversifies their CRE exposure and generates fee income. The AI narrative is not a cure for office distress. It is a distraction that lets bank executives tell investors they have something better to do than worry about 2019-vintage office loans.
But the incentive map is not uniform. The big banks cut reserves because they have the balance sheet, the diversified loan book, and the fee income to absorb the losses they already recognized. Their office exposure is a smaller share of total loans than it is for regional banks. Their cost of capital is lower. Their ability to hold nonperforming assets without triggering regulatory scrutiny is greater. The banks that reported Tuesday are not the banks that will define the next phase of the office distress cycle. The regional banks, the debt funds, and the CMBS special servicers are the ones whose constraints are about to be tested.
Consider the mechanism. The big banks cut PCL because their models say the losses are already priced into the loans they hold. That does not mean the losses are zero. It means the losses are known. For a regional bank with a concentrated office portfolio, the same model might produce a different answer because the portfolio is less diversified, the collateral is lower quality, and the refinancing options are fewer. The big banks can afford to mark their office loans to a realistic value and move on. Regional banks cannot, because marking to market would erode capital ratios and trigger regulatory attention. That is why the office distress cycle will play out in slow motion on regional bank balance sheets, not in the quarterly earnings of the money-center banks.
The practical implication for CRE owners and lenders is straightforward. If you have an office loan at a money-center bank, the bank has likely already decided whether it will extend, restructure, or foreclose. The reserve cut suggests the bank believes it has enough capital to cover the losses it expects. That does not mean the loan will be renewed. It means the bank is not expecting a surprise. If you have an office loan at a regional bank, the uncertainty is higher. The regional bank has not yet recognized the losses, and it may not have the capital to absorb them without raising equity or selling assets. The clock is different.
The AI and data center buildout is real, but it is not a substitute for office demand. It is a separate capital flow that happens to be large enough to absorb some of the construction lending capacity that banks pulled back from office and retail. It does not put tenants back into 2019-vintage office towers. It does not lower the vacancy rate in Class B buildings. It does not make the 2027 maturity wall easier to refinance. What it does is give bank executives a credible story to tell investors about where the growth is coming from. That story is working. The stock market rewarded the earnings reports. But the story does not change the math on a 15-year-old office building in a secondary market with 30% vacancy and a loan coming due next year.
The next thing to test is the regional bank earnings season. If the regional banks also cut reserves, the market will interpret that as confirmation that the office cycle is truly behind us. If they hold reserves flat or increase them, the market will read it as a signal that the losses are still coming. The big banks have already taken their medicine. The regional banks are still deciding whether to swallow it.
The banks are not saying office is fine. They are saying office is priced in. That is a meaningful distinction. Priced in means the losses are known, the capital is allocated, and the market can move on. It does not mean the losses are small. It means they are no longer a surprise. For the owners of office assets that are not yet marked to market, the surprise is still ahead.