Strategic Decisions

Bernstein's 42nd Annual Strategic Decisions Conference was held in NYC this past week. AI, the economy, and stablecoins were three topics discussed amongst many. Here's what those conversations sounded like.

“The economy seems ok, right?”

On the surface, the economy, right now, feels comfortable. Unemployment is low. Consumer spending is holding. Credit losses are benign. The problem is that the most experienced people in that room on Thursday and Friday know that comfortable surfaces can be deceptive.

Jamie Dimon opened his session by pushing back hard on the idea of running a single base case: "The One Big Beautiful Bill is $300 billion of stimulus. Deregulation is a form of stimulus. AI expenditure is a form of stimulus — $300 billion year-over-year, another $300 billion for next year. Money supply is going up. Banks lending more money. That's what we have today." His conclusion wasn't that disaster is imminent. It was that the range of outcomes is wider than markets are pricing, and anyone who tells you otherwise is giving you false comfort.

Bank of America's Brian Moynihan offered the most granular consumer read of the conference. In the first 20 days of May alone, BofA customers had moved money out of their accounts at a 5% pace above last year — consistent with what he's seen across Q1 and April. "People are spending money, and that's because, frankly, they're employed. When they're employed and earning money, they're in pretty good financial position." But he was careful to point out the tension hiding beneath that strength: small and mid-sized businesses are being deliberate about hiring as they reckon with uncertainty around trade, oil prices, and AI's impact on headcount. The consumer engine is running, but it's being watched closely.

Apollo's President Jim Zelter described the environment with a golf analogy that has stuck with me: "In the middle of the fairway, things are nice — massive CapEx cycle, big M&A cycle, risk-on equity mentality, good corporate profits. The challenge is, in the rough: inflation, global fragmentation, the crisis in the Middle East. The problem is that fairway is getting narrower, and it feels like it's the narrowest fairway we've had in a long time." Apollo's response has been to lean hard into High-Grade Capital Solutions — investment-grade companies in the industrial renaissance — which Zelter described as carrying their deepest pipeline in company history.

Carlyle's Harvey Schwartz took the longest view. He argued the geopolitical splintering we're living through is a structural regime shift that's been building since Russia went into Ukraine. "National security's been completely redefined. It used to be narrowly defined as defense. Now it's defined as energy, defense, data." His contrarian read: less available capital globally, higher demand for it, and therefore better marginal returns for firms positioned in the right places. For Carlyle, that means aerospace, defense, healthcare, and industrials — all of which have gone from "old economy" to "exactly where the world needs capital" almost overnight.

“AI is an accessible, transformation tool, not a competitive advantage.”

Every CEO on stage mentioned AI. But the most valuable contributions weren't the bullish ones — they were the honest ones about what AI can and can't do.

Dimon again: "I don't agree with this notion that if JPMorgan does it first, we're going to create higher margins and that lasts forever. It lasts temporarily. What happens in a competitive world is if I do something better, well, so is everyone else eventually, and that gets competed away." 150,000 JPMorgan employees use LLMs weekly. The savings are real. The moat is not.

Datadog CEO Olivier Pomel had the sharpest technical take of the entire conference. When asked whether AI agents would eventually replace what Datadog does — a bear case that circulates constantly — he didn't dodge it. He reframed it: "The companies that are building these kinds of models are also using our product to manage their infrastructure. It is a hard job. It is consistently a harder job than writing the application itself." Then came the analogy worth writing down: "You could upload a photo from your dashboard to Claude and ask what you should do, and it will give you a pretty good answer. The problem is that Claude will not drive your car in real time. If it did, it would do it poorly — and it would cost you $10,000 a day." Probabilistic AI and deterministic real-time infrastructure are complements, not substitutes. Companies that understand that distinction will make much smarter capital allocation decisions than those treating AI as one unified force.

ICE founder Jeff Sprecher, one of the more unconventional voices at the conference, connected AI directly to the future of data monetization. He acknowledged that investors are uncertain about how AI will change how data is consumed and paid for — which he sees as a misread. "We think those are tailwinds for us, not headwinds." ICE has been early in building AI models both internally and with customers, and his argument was simple: more supply chain complexity, more geopolitical rewiring, more participants needing to hedge — all of it flows to their markets and data businesses.

Klarna CEO Sebastian Siemiatkowski brought a different angle — not about AI reducing costs, but about AI enabling a business model that incumbents simply can't replicate: "We have introduced... concepts that some of us remember from when they were younger, that banks abandoned." His broader point was that Klarna's edge isn't AI per se — it's 20 years of consumer behavior data and global market experience that AI can now be applied on top of to build products that traditional banks structurally won't build because those products reduce revolving credit balances.

Will stablecoins unlock internet commerce?

Stablecoins and tokenized assets were discussed throughout the conference.

ICE's Sprecher was probably the most direct on where this ends up: "We're going to exchange value over the internet, and it will be tokenized and encrypted... I think as we move to 24-by-7, 365 trading of things around the world, the capital is going to move on the internet." He's already working with three large banks tokenizing deposits, and his view is that stablecoins — for now a retail product — will ultimately just become a faster, cheaper set of wires moving money that already exists in institutional accounts.

Mastercard CEO Michael Miebach connected stablecoins specifically to the next wave of commerce: agent-to-agent micro-transactions. He described Mastercard's Agent Pay capability as the trust layer for agentic commerce — established protocols with Microsoft, OpenAI, and Google, available to issuers globally. Then he was precise about the stablecoin fit: "If this is higher frequency, instant micro-transactions that are the fraction of a dollar, our stablecoin capability may just be about the right thing to go after those additional transactions." He cited the BVNK acquisition as the infrastructure play behind it.

Visa CEO Ryan McInerney pushed back on the premise that stablecoins threaten the card network — but in a way that actually revealed how seriously Visa is taking the question. His argument: "In an agentic world where you're going to send your agent out to buy things for you, what are you going to want to use? Are you going to want to use a credential that has broad-based acceptance, that has well-established rules, where if you get the wrong item you know you can return it and get your money back? Compare that to stablecoins — an instant, irrevocable transaction." Visa's play is that trust, fraud protection, and acceptance are the moat — and they're rebuilding product development from the ground up to make sure those moats hold in an agentic world.

Bank of America's Moynihan made the institutional concession simply and directly, in the middle of an answer about competitive positioning: "We look hard at payment systems changes, obviously stablecoins and the legislation going through and our positioning, both as a company and industry, because that's a big driver of profit." That's a CEO of a $3.5 trillion bank putting stablecoin legislation in the same sentence as profit drivers. That's where this conversation has arrived.

Figure Technology's Michael Tannenbaum put the arc of the shift into context: "Two years ago, tokenization was not in the lexicon. Today at a bank board meeting, there are conversations about tokenized deposits and stablecoins in a way that you would've been laughed out of a board meeting previously. Now people are asking, 'What's our strategy?'"

The common theme throughout

What I keep coming back to is how consistent the best thinkers in that room were about one posture: humility about the limits of their own advantage. Dimon said JPMorgan's AI edge is temporary. Pomel said the AI bears are asking the right questions even if their conclusions are wrong. Sprecher said his company is misunderstood, the market will eventually discover its value, until then they go back to work. Moynihan tracked his data at the transaction level and refused to extrapolate beyond what it actually showed.

These discussions and comments are a strong reminder that even the smartest in the room work extremely hard to stay rigorously grounded in what they actually know and deliberately positioned for the range of possibilities of what might comes next.

#Investing #PrivateEquity #AI #Stablecoins #Macro #BernsteinSDC #JPMorgan #BankofAmerica #Apollo #Carlyle #Visa #Mastercard #Klarna #Datadog #ICE