The Only Metric That Matters

I round up the most relevant AI-in-finance news - the deals being done, who’s rolling out what, and what’s actually working on the front lines

…A fintech called Altruist launched an AI tool that generates optimised tax strategies in under three minutes. Raymond James fell 8.8%. Schwab dropped 7.4%. More than $20 billion was wiped from wealth management stocks in a single session. Bloomberg ran the headline: "AI Threatens the Finance Industry's Perpetual Profit Machine."

Anthropic closed a $30 billion Series G on February 12 at a $380 billion valuation. Run-rate revenue has hit $14 billion. Goldman's David Solomon stepped in to call the software selloff "a little bit too broad." McKinsey declared that AI M&A has entered its "industrial phase." US deal activity rocketed 111% in January.

But first, my take on what all of this means for deal economics. Because I think the market is still dramatically underpricing it.

In This Week’s Issue:

From The Trenches

The Only Metric That Matters

A report from Apollo caught my eye last week. "Private Equity Returns to Its Roots." The thesis: the easy money era is over, and returns will increasingly come from pricing discipline, asset selection, sector expertise, and operational value creation. Standard stuff. Well-argued.

What struck me was what wasn't in it. AI barely featured. In a piece about what will drive PE returns going forward, the single most important variable was essentially a footnote.

And this in a week where a single AI product launch from a fintech most people hadn't heard of wiped $20 billion from wealth management stocks. The FT is calling it private equity's "doomsdAI moment." Dan Primack at Axios pointed out that VC and PE firms have been touting how much work can now be done by AI, which simultaneously undermines their own value proposition. And yet the industry's own thought leadership still treats AI as a sidebar.

I've been saying since December that the timeline for meaningful disruption to white-collar work is 12 to 18 months. Each week the evidence gets harder to dismiss. Everyone is massively underpricing this, even as it knocks on the front door.

Most people still think of AI as an efficiency play. Fewer analysts, faster models, cheaper back office. That's relevant, but it's only one part of the story.

Start with sourcing. AI lets you identify and understand better signals of intent across the market. You're not waiting for a banker to send you a book. You're picking up on patterns, monitoring triggers, capitalising on opportunities before they hit the market. The top of the funnel widens dramatically.

Then think about what happens when those opportunities cross your desk. AI compresses a 100-day diligence process into a few days. Not with a team of 5 deal professionals and 10+ external consultants, but with 2 people and 10 agents running the whole thing. The marginal cost of evaluating the next deal approaches zero.

But it's not just the evaluation. It's the full underwriting of the opportunity. Previously, bolt-ons would get a day-one uplift from margin benefits of a few percentage points. The gap was going from 12% margins to 15%. Now the gap is going to 30%, 45%, 50%, because the cost structure can be so dramatically improved through AI.

And that's only the efficiency side. The more underappreciated angle is revenue. Everyone focuses on cost-cutting. But AI lets you scale without proportional headcount growth. It becomes far more viable to enter new markets, launch new products, reach new customers, expand service offerings. The ceiling on what a well-run company can achieve rises by an order of magnitude.

Deals that were previously uneconomic to evaluate. Too small, too complex, too many. They suddenly become viable. The total addressable universe of deals expands.

That's why entry pricing starts to matter less. Apollo argues pricing discipline is critical when the easy money era ends. But if the best operators can dramatically expand both margins and revenue through AI-enabled growth, the difference between paying 8x and 12x starts to shrink. The value creation available to the firms that move first dwarfs the entry multiple premium.

The real implication isn't about being right with your picks. It's about rewarding the firms that move fastest. Because those who get ahead are going to run away with the market.

Every industry is about to become a race. AI removes the constraint that kept companies in their lane. You're going to see accounting firms become advisory firms, then consultants, then wealth managers. You're going to see mid-market platforms use AI to broaden their offerings and scale into adjacent categories. The opportunity to do more with the same infrastructure explodes.

The result is monolithic winners. A few firms in each sector that move first, scale fastest, and consolidate everything around them. This is winner-takes-all dynamics, and the only way to get there is to race.

That creates a problem for traditional PE exits. The buyer pool shrinks. The middle market gets hollowed out. So the goal shifts. Instead of buying, improving, and selling, the playbook becomes: race to establish one of the winning positions and figure out the return of capital later. Dividends, recaps, other structures.

PE starts to look less like a buy-and-sell model and more like a buy-and-hold-and-compound model. I'm going to come back to this in a future issue, because the implications for fund structures, LP expectations, and exit timelines are worth unpacking properly.

News Digest

McKinsey: AI M&A Enters Its Industrial Phase

McKinsey's latest M&A report makes it official: AI dealmaking has moved past the experimentation phase. Global M&A value hit $4.7 trillion in 2025, up 43% year-over-year and 20% above the ten-year average. Technology, media, and telecoms accounted for 23% of all deal value, growing 61% to $1.1 trillion.

The numbers within tech tell the real story. AI-native companies are commanding the highest multiples on a per-deal basis. Semiconductor consolidation and computing platform roll-ups are accelerating. And the new M&A playbook, according to McKinsey, is shifting from traditional scale economics to access to talent, proprietary data, and model IP.

The details:

Why it matters: The acquirers are no longer buying AI companies for optionality. They're buying them because the business model requires it. That's what "industrial phase" means.

My take: The stat that jumped out: 40% of respondents using gen AI in M&A reported 30-50% faster deal cycles. That's today, with early-stage tools. Now extrapolate 18 months. The firms building AI into their deal processes right now are accumulating a compounding advantage that will be very difficult to replicate. And as I argued above, the economics of what those firms can achieve with their portfolios are changing too. The premium on AI-native targets isn't for the revenue. It's for the capability.

Read more

Wall Street Tries to Calm the Software Selloff

Goldman Sachs CEO David Solomon stepped up to the microphone at a UBS conference in Key Biscayne last week and delivered a message to rattled investors: the software selloff has been "a little bit too broad." There will be winners and losers, he argued, and plenty of companies will pivot and do just fine.

He wasn't alone. Across Wall Street, senior executives have been working to steady nerves after a week that saw more than $300 billion wiped from software and data stocks. Shares in Apollo, Ares, Blackstone, and KKR all came under pressure given their known exposure to software companies threatened by AI. The FT reported that PE's big bet on software was being "derailed by AI," with UBS estimating 25-35% of private credit exposed to disruption risk.

But the real damage is showing up in deal flow. Bankers say the selloff is now disrupting M&A and IPO pipelines in the sector. Volatility has made valuations unreliable and buyers cautious.

The details:

Why it matters: The people running the biggest firms in finance are telling investors to calm down. But they're saying it from podiums at sunny Florida conferences while their own portfolio companies are getting repriced in real time.

My take: The fact that the selloff is already freezing deal activity tells you the market isn't just spooked. It's recalibrating. Solomon is probably right that the selloff was indiscriminate. Not every software company is equally exposed. But "a little bit too broad" isn't the same as wrong. The direction of travel is clear. The firms that steered clear of software, like Apollo itself, are outperforming. The ones that loaded up are scrambling.

Read more

Anthropic's $30B Raise and the Pentagon Standoff

Anthropic closed the second-largest venture deal in history on February 12. $30 billion in Series G funding at a $380 billion post-money valuation, more than doubling from $183 billion. Led by GIC and Coatue, with 36 investors including D.E. Shaw, Founders Fund, Microsoft, and Nvidia.

The numbers tell the story. Run-rate revenue has hit $14 billion, growing 10x over three years. Claude Code alone is at $2.5 billion, having doubled since January.

But the more interesting development happened the same week. The Wall Street Journal reported that the Pentagon used Claude during the military operation to seize Venezuela's Nicolás Maduro, deployed through Anthropic's partnership with Palantir. Anthropic responded by threatening to void its $200 million defence contract. The Pentagon is now pushing AI labs to allow their models to be used for "all lawful purposes." Anthropic insists two areas remain off limits: mass surveillance of Americans and fully autonomous weaponry.

The details:

Why it matters: The company that just raised $30 billion is simultaneously picking a fight with the US military over where to draw the line. That tension between scale and principle is going to define how enterprise AI develops.

My take: Anthropic's willingness to walk away from $200 million tells you something about how they're thinking about the next decade. They're building for enterprise trust, not government contracts. That's a bet that the commercial market, banking, PE, legal, healthcare, will be worth far more than defence. Based on the Goldman partnership from last week, they're probably right. The firms evaluating AI vendors right now are paying attention to which companies draw lines and which ones don't.

Read more

Other Interesting Things I’ve Read of Seen This Week:

US M&A activity rockets 111% as megadeals return (Feb 12) - January deal volume more than doubled year-over-year. European M&A also ignited with $60 billion in February alone. Everyone's deploying capital again. The question is whether their underwriting accounts for what AI is about to do to the targets.

PitchBook launches first daily valuation model for VC-backed companies (Feb 12) - Real-time marks on private companies. LPs everywhere just felt a chill run down their spine.

ElevenLabs raises $500M at $11B valuation (Feb 10) - Voice AI startup that didn't exist three years ago is now worth more than most of the software companies getting hammered this week. Timing is everything.

Former GitHub CEO raises $60M for startup Entire (Feb 10) - Thomas Dohmke's thesis: a future where "humans no longer look at computer code." Bold claim from the man who ran the world's largest code repository for four years.

PE firms profit by flipping gas plants to AI-hungry producers (Feb 11) - While PE's software bets are getting hammered, their energy infrastructure plays are printing money. AI needs power. Power needs gas. PE owns the gas plants. Sometimes the best AI trade is the dumbest one.

Cisco's president says AI agents need "background checks" like human workers (Feb 13) - Jeetu Patel wants identity verification, access controls, and audit trails for autonomous agents. Basically, HR for robots. He's not wrong.

Acquisition Intelligence is a weekly newsletter on AI in M&A for finance professionals, private equity investors, investment bankers, corp dev teams, and deal-makers.

For questions, feedback, or to share what you're seeing in the market, reply to this email.

P.S. I'm Harry, co-founder of DealSage. We're building an AI-native deal intelligence platform to help professionals turn their institutional knowledge into better decisions. If you're curious what we're up to, check out dealsage.io or just reply here