Bank of America told investors its trading revenue will be flat this quarter, and its chief executive says that would still be one of the best third quarters in the firm's history.
A flat print usually means something has gone wrong. Brian Moynihan's argument is that the comparison is the problem rather than the business: the second quarter was explosive, the AI trade has come out of the system, and rates moving 100 basis points in a short stretch made issuers wait.
"It's just sometimes the market quits doing things, and that's one of these times."
Moynihan has run Bank of America since 2010, bought Merrill Lynch in the week Lehman Brothers failed, and spends more than $400 million a year deploying AI inside the bank.
The full segment is covered here so you can skip it.
Here are the 8 calls that matter.
👤 Guest: Brian Moynihan, CEO of Bank of America
🎙️ Host: Daniel Berger of Bloomberg TV, interviewing him on stage at the Barclays Global Financial Services Conference in New York
📰 Published: 14 September 2026 on the Bloomberg Businessweek Daily podcast
🟣 Apple Podcasts | 🔗 Episode page | ⏱️ length not available
Key Takeaways
Flat trading this quarter would still rank among the best third quarters the bank has ever had
Equities are up, fixed income is slightly down, and the AI-trade leverage has come out of the system
The bank will not run fully autonomous agents, because it cannot see how it would control them
It spends more than $400 million a year on AI and plans to spend more next year
Moynihan had already called for an AI slowdown before the AI companies did
His regulatory ask is a single federal rule rather than fifty state ones
He says banks regulate themselves more tightly than their regulators do
A slowdown in frontier models would not slow the bank's own AI gains, because it does not use frontier models
The versions in production are a generation behind the ones being argued about
He still expects an AI super cycle, on the reasoning that customers are willing to pay
His team has forecast three Federal Reserve rate increases and does not see inflation at target until 2027 or 2028
He reads a 5% 10-year yield as a return to normal rather than a stress
Consumer spending was up 4% in August year over year, and credit quality is strong
1. Flat Is a Record Quarter
Berger opened on the news Moynihan had made minutes earlier on the conference stage — guidance that trading revenue will be flat this quarter — and asked what had changed, given how large the second quarter had been.
The first move was to reject the framing: "So I think just to keep a perspective, if we end up with a quarter of what we just said flat, it'll be one of the best third quarters that we've ever had in the trading history of the company."
The comparison is doing the work. Last year's second quarter, he said, was the tariff quarter, and last year's third quarter was the recovery from it — a period when, as he put it, everyone was running around saying the economy was going to stop cold and then found it easier than feared
This year's second quarter had a different driver, and it has gone: "This year, then the AI trade came in, and you guys report about Korea and the leverage and all that stuff, and a lot of it's come out of the system because people brought the risk down."
The mix underneath the flat number is split: "Now, equities is up more, and fixed income is bouncing around a little bit down, and that ends up to flattish."
"It's just sometimes the market quits doing things, and that's one of these times," he said of his markets business
Investment banking is the opposite case — a full pipeline held back by the rate move: "Now, the pipeline investment banking's full. The activity level is strong, but with rates moving up 100 basis points over a short period of time, people pause a little bit, and leverage coming out of the system for a while."
2. What Comes Back in Q4
Asked whether the pause runs through the fourth quarter, Moynihan made the answer conditional on the rate path rather than on client demand.
The unlock he named is stability, not direction: "I think as soon as we get some stability around the rate structure, I mean, they got the Fed meeting this week, and sort of what people get a common view of that, then the rates will settle at some point, and I think that'll help some of the trading activity."
The equity calendar is already committed: "We've got a lot of pipeline IPOs, secondary offerings, things coming on that we know will go through." Debt financing is the piece that needs a settled rate structure, because an issuer has to be able to commit to a deal and still price it days later
He put the flattish quarter at more than $5 billion of revenue and called it good, and described the bank as 60% core net interest income — the spread between what it earns on loans and pays on deposits — which he said is doing everything the bank expected
The rest of the franchise is not the problem: "The credit quality of the firm is good. The consumers are spending well. It's just that the trading activity was such explosive in the second quarter." When it settles down, he said, it feels strange, and then it grows from there
He also put part of it down to the calendar: "And you've got the summer's a summer, and we lose a lot of activity in August anyway." A second- or third-quarter decline, on his account, is traditional
On the fund Situational Awareness, for which Bank of America acted as prime broker, Berger asked what had changed after the blow-up. Moynihan declined to discuss clients: "But you should assume that we have good risk manager practices then and now." The bank, he said, made money every day through it, but that does not mean the environment cannot change quickly
3. Should They Slow Down?
Berger noted that Moynihan had said on Bloomberg in July that he was structurally worried about AI's impact, and asked whether he is worried about where this is heading now that the conversation has moved to humanity itself.
His answer was that he had already given it, on another network the week before: "And I said, I think they should slow down, but I'm not the person running these companies. They have now said they should slow down. So I think that's interesting."
The image he reached for was a race with its own drivers asking for a speed limit: "You basically had the analogy I said to somebody, we were talking to somebody today, was you have the NASCAR drivers saying, please slow down the cars." Who decides how, he said, is the open question — the governing body, meaning Congress or the administration; the private sector refusing to buy a product with those impacts; or society at large
The specific line he would not cross is autonomy: "That's far different than using AI in a controlled environment. When you put a completely autonomous agent out there, that's the big concern."
He described two failure modes he is watching — an agent used for hacking, and agents coordinating against people — and said the second is what has been in the papers
He read the companies' move as self-interested and sensible: "So I think they're wiser saying, hey, we have a great product here. Let's not gum it up by having to do some things that cause an over-regulation."
4. $400M a Year on AI
Berger's next question was whether the financial industry has a role in how AI develops, given that it finances the build-out even though it does not build anything.
Moynihan's first answer was that the bank is a customer before it is a financier: if these companies are going to have a revenue stream, it comes from companies like his
Bank of America has deployed more than $400 million in AI capabilities this year, he said, and will deploy more than that next year
The conviction is old rather than new: "It works. Our Erica we developed 10 years ago. So we believe in the power of digital manipulation of data and information and process. Believe it to our core."
The constraint is accuracy, and he made it a customer-service point: "If you went into Erica and used it and we gave you the wrong answer, not a great client experience." That is why the bank deploys with controls and people in the loop
The financing question he framed as a liability question: "As a financer, we have to worry about what's the liability." There is a lot of talk about product liability, he said, and a bank has to think about it as a lender while the markets think about it as a funder
The precedent he cited runs slower than the technology: "And product liability around products from cars and things took years to develop, so the question is how you do it."
He added that a competitive race with other parts of the world adds a complexity the bank cannot control, and said Bank of America works with the AI companies regularly on hacking and cybersecurity as well as on applications
5. Who Regulates the Agents
Berger asked whether AI risk belongs in an IPO prospectus and, separately, whether regulators should step in given the administration's opposition to guardrails.
On the prospectus, Moynihan pointed at regulation as the risk factor that matters: if the issues got bad and the response were over-regulation, the revenue stream would arrive more slowly. He noted he had been a lawyer himself and assumed the lawyers were working through the risk factors
His answer on who regulates was unambiguous: "Well, I think it's always a self-regulatory moment. Because in the end of the day, you shouldn't require, just like with us regulation as a bank, we regulate ourselves tighter than the regulators do."
He dated the argument to his own experience: "Here we are, really on the anniversary of Lehman failing and us buying Merrill Lynch at Bank of America in just a hugely disastrous time for the financial services industry, from which we recovered and done a great job. That was a lot of self-regulation, honestly, and that's important."
What he does not want is fifty different rules: "What we advocate for is, wait, if you have everybody regulated, then you have a patchwork of stuff that's hard to figure out. So can we operate with our tools and capabilities in California, same way you can operate in Nevada, same way you operate in Texas and in North Carolina." He named the plaintiffs' bar, state attorneys general and state legislatures as the sources of that patchwork, and said the answer is federal regulation
The rule he applies inside his own firm is stricter than anything being proposed: the bank would not let itself deploy autonomous agents, because it cannot see how it would control something whose job is to go and solve problems over and over again
Semi-autonomous agents are a different matter, and he called the work around them pretty interesting stuff
"But it's good news that the NASCAR drivers are saying, let's slow down the engines and make sure we can be safer," he said, closing the thread
6. Still a Super Cycle
Berger put the commercial question directly: banking has done extremely well out of AI through trading, mergers and initial public offerings, so what does a real slowdown cost the industry?
Moynihan's answer is that the bank's own AI value does not depend on the models being argued about: "The models we're using now to get all the value we're getting at Bank of America and other companies are not these models. These are the frontier. These are the most advanced. The models we're using are versions before that. These models haven't even been deployed yet. This is in a test."
So the deployment, and the data centers to run it, happen either way in his account — the only question is the shape of the growth curve, and he acknowledged real debate about whether it can be built as fast as anyone wants
The conclusion is a demand argument rather than a technology one: "It'll be a super cycle, I'm fairly convinced, because we see the value. It means we're willing to pay something for it." A customer willing to pay creates a revenue stream that builds, and whether it turns out to be 75% or 50% of the projected size, he said, it is still a lot
7. Missing Top Billing
Berger asked how much of the merger cycle is AI-driven, and why Bank of America does not always get top billing on the biggest listings, naming SpaceX as an example.
Moynihan put it down to the length of the relationship, and said the bank did participate in SpaceX
The distinctive asset he named is distribution to individual investors, which he said matters for two separate reasons: "A lot of these companies want to get out to the broad base for two reasons. One is a good shareholder base, but secondly, it's also a good user base." Retail holders, on his account, are the people who keep product knowledge circulating
The lumpiness is a timing artifact rather than a share loss: a transaction pushed into the next quarter moves the fee with it, because the reported number is cash received rather than deals announced
He gave the underlying run rate and said it is a good quarter: "But the underlying flow of business, 1.6 to 1.8 billion, which is what we told people, that's a good strong quarter." He added that two-billion quarters are rare — the bank had one in the same quarter last year, which he called an even better quarter
He described the ambition as integrating a corporate and investment banking business inside a franchise that makes a couple of billion dollars a quarter
8. 5% Is Normal, Not High
With a minute left, Berger raised the rate decision due that Wednesday and a 10-year Treasury yield that had reached 5% for the first time since 2023.
Moynihan separated the two ends of the curve and said the bank's house view on short rates has been unchanged for a long time: "And I think the short-term yields, the Fed raising rates, our team has had the Fed doing three rate increases for a long time now." He said the market had not agreed and that it now looks closer to being right
Three rises, whether this year or two now and one early next year, are about unwinding the cuts already made — and even then he does not expect success soon: "We don't think inflation, even with three rate rises, would get down to where the target would be until the end of 27 and 28."
The evidence he offered that the economy can take it is his own customer data: "And the economy is robust. Our consumers spent 4% more money in August than they did last August. Their credit quality is very strong. The small, medium-sized businesses are borrowing money."
The transmission runs through businesses before households, which he said is counter-intuitive: "Interestingly enough, short-term rate moves affect the small, medium-sized line of credit users faster than they do the consumer because mortgages are fixed, cars are fixed, and credit cards, the rate move doesn't mean as much."
He said the bank has not seen stress from the move, on the grounds that rates were at these levels a year to eighteen months ago and borrowers were fine
His framing of the level itself is the argument the segment ends on: "And then the rate structure can be more normal. Everybody says higher. It's actually more normal."
"It was not normal from 2009 to 2019 to have zero rates. That is not normal," he said, calling a normalized rate structure a good place for the US to be and one that helps the rest of the world normalize too
Bonus Insights
Moynihan said he is attending a meeting convened by the King in Scotland in his role with the Sustainable Markets Initiative, which he has worked on with the King for seven or eight years
He did not have the guest list, and said the convening power is the point: he has seen the same thing work on the energy transition and other subjects
His list of who is now engaged with AI risk ran well beyond markets — the Pope has written on it, the King has convened people, the companies themselves have moved and the administration has thought about it
Berger noted that Jensen Huang is expected at the same meeting
Moynihan's bottom line is that Bank of America is a large, paying, deliberately conservative AI customer — spending more than $400 million a year on models a generation behind the frontier while refusing autonomous agents outright — and that the rate structure everyone is calling high is the one he would call normal.
Products, Companies & Tools Mentioned
Bank of America (Moynihan's firm; flat trading guidance, a full investment banking pipeline, more than $400 million a year deployed on AI and a book that is 60% core net interest income)
Erica (The bank's virtual assistant, built ten years ago, and Moynihan's example of why an AI deployment has to give the right answer)
Merrill Lynch and Lehman Brothers (The 2008 purchase and the failure he dates the bank's self-regulation argument to, on the anniversary of both)
SpaceX (Named by Berger as a listing where Bank of America did not take top billing; Moynihan said the bank did participate)
Situational Awareness (The fund Bank of America acted as prime broker for; Moynihan would not discuss the client but said the bank made money every day through the episode)
Sustainable Markets Initiative (The body Moynihan represents at the King's meeting in Scotland, after seven or eight years working on it)
Federal Reserve (Bank of America's house view is three rate increases, with inflation not reaching target until 2027 or 2028)
If this was worth your time, send it to someone who follows the name.
Get the latest market chatter as it happens: