They Paid $7.7 Billion for a Broker — and Intend to Do One Deal a Year

Sequence Holdings and the Dell Family Office have taken insurance broker Baldwin private for $7.7 billion, a deal host Sarah Guo calls the largest AI take-private to date. Michael Lee, Sequence’s co-founder and CEO, tells her the firm is a permanent holding company that partners with management teams to buy businesses and re-found them as market leaders with AI. The brokerage space has been on the firm’s list since founding. After meeting dozens of brokers, he says Baldwin was “end of one,” and the coming years will be a shared journey among Sequence, Trevor Baldwin’s team, and Dell.

The operating cadence is as unusual as the check size. Sequence is not an investment shop trying to put money to work ten times a year. Lee says the work is scarce: one deal per year, and if that means none, that is fine. Guo is not a distant interviewer. She wrote Sequence’s first check, and Lee is a career investor, not a serial founder — Goldman out of college, Apollo for private equity, then Lone Pine from 2017, where he covered public markets and later built the firm’s private investing business. The bet is that a professional investor can own an old giant and re-found it.

When Lee joined Lone Pine in 2017, one of the first books he was asked to cover was AI. AlphaGo had just landed, the first Transformer paper was out, and the commercial question was whether convolutional nets, GANs, and LSTMs would become businesses if you scaled them. Exciting demos arrived, he recalls, but those architectures degraded as you pushed them. That history is why late 2022 felt different. ChatGPT, in his view, meant the world had forever changed: for the first time there was an architecture people knew could scale infinitely with compute, even if nobody could say what would come of it.

Thinking like an investor, he expected an uneven shock. Restaurants and golf courses, he guessed, would barely feel AI. Coding was the opposite: a startup would win, and buying an outsourced coding shop would be a terrible idea — the money should go to firms such as Cognition or Anthropic. Then there was a third bucket, the one that kept him up. As a student of business, he had always believed some industries leave brand, scale, network effects, and regulation in the incumbent’s hands. In the biggest technological shift of a lifetime, could you acquire the right one, inherit those advantages, and forge the market leader?

That kernel showed up in early 2023. A handful of people tried to pull him out of Lone Pine to go do it then; it was not the right time. In 2024, on a walk with his wife, he said it is rare to be in the middle of the most important technological change of your lifetime, to have an idea you believe in, and to feel you can assemble the team. He and co-founder Alex started Sequence. Guo notes that many venture investors see incumbent advantages and walk the other way. Lee’s reply is that the model meets the moment: this is less ambition than a job that has to happen, and he does not see a natural way for it to happen unless you own the company.

Sequence treats the moment as the next industrial revolution. What Lee sees inside most enterprises, and from software companies shipping agents, is a push to hand a small machine to every person on the human assembly line and speed up work. That is better than nothing. But when machines can run 24/7, scale with electricity, and do what no person or team can, the real question is how the organization itself should be rebuilt. A holding company, in his telling, is the only structure that lines up duration of capital with the investment required, orients culture around forging market leaders rather than deploying money, and lets retained earnings fund both existing companies and new ones.

He walks through a thought experiment. You are CEO of a Fortune 500 company, charged with re-founding yourself. Option one is to do it in-house. The structural block is talent: every company has a celebrated persona, and that persona decides who gets hired and who stays. Blackstone celebrates the investor, which is why it aggregates great investors — Lee even has every Sequence employee read *King of Capital*. Palantir, as he describes it, aggregates world-class engineers, celebrates them, and wraps that capability for organizations that cannot get the talent. An insurer or a bank celebrates sales and risk. Asking it to switch personas on a short clock, he argues, is not realistic.

Option two is the multi-trillion-dollar services industry — Accenture, McKinsey, Palantir. Those firms do celebrate engineers and technology change. The problem is incentives. Services companies, Lee says, optimize for getting in your wallet, staying in your wallet, and growing their share of it. That path is incrementalism. Changing how people are organized, who works there, and how they are paid would, if done fully, put the vendor out of the account. Option three is buying software. The obvious flaw is that a tool available to everyone is not an advantage. The subtler flaw is genetic: a software company hunts for a homogeneous workflow, installs it fast and deep, and sells to the assembly line as it exists today. You cannot sell a product around a new line that does not exist yet.

Ownership is the remaining door. Alignment, a culture that celebrates the engineer, and a long horizon let Sequence attempt the surgery software vendors will not sell and consultants will not recommend. Early evidence, he says, is already on the ground at BankSouth.

Guo admits that when Lee first said Sequence would buy a piece of a bank, her reaction was mild horror: banking is not a popular private-equity sector, and for reasons. The origin story was a cold start. Without money you cannot do a deal; without a deal you cannot hire engineers; without engineers nobody wants to give you money. Guo’s check and a handful of engineers still left them with nothing to work on. Buying any asset just to prove they could, or waiting for the perfect one, both looked wrong. They took a customer instead. Lee called Jamie Reynolds, a co-founder of Avenir, whose family owns a community bank in Georgia. Work started on August 4 — Lee’s anniversary, which he skipped.

From August through late November they operated as a services firm, attacking the bank’s key workflows. Then the family asked whether they would become a permanent partner. That became Sequence’s first investment, closed in March after Fed and OCC approval. The regulated setting, Lee argues, was a feature. How the institution operates is well defined, data hygiene is excellent, and the rules are clear — an environment that works unusually well for agents. He also talks about organizational physics. BankSouth has more than $100 million of sales and many branches, but underwriting sits at headquarters, so anything built there amortizes across the network. That is a different complexity class from a roll-up that must integrate systems, send engineers to many sites, and standardize procedures and culture.

The numbers from the first six months after the March close are the hardest facts in the conversation. Sequence built a system that now takes on all consumer loans; average consumer underwriting work is down 94%. Commercial underwriting has started to roll out, and the average loan that used to take 30 days end to end now takes 11. In the second quarter, loan volumes doubled versus the first — luck, Lee insists, not Sequence’s doing. Historically the bank would have turned business away because the middle and back office had bandwidth limits. This time it absorbed the volume without changing underwriting standards, and with a smaller underwriting team: one person retired, one moved to the front office.

What he wants from that is not a headcount story. Loan officers spend more time in the field and less time writing credit memos. Underwriters work the hardest exceptions instead of deciding whether to issue a boat loan to someone who clearly has the money. People, in his phrase, get to do what they are best at, and the bank can take on more volume with the same or fewer people in the middle. Looking ahead, he wants new products, better tools for relationship managers, and a customer experience no one else in the industry can match.

Sequence does not scale by doing ten deals a year, so each one has to count. Lee wants companies that can become $100 billion-plus businesses, in industries where the incumbent holds the advantages, and where what the organization does maps onto what AI is good at today. Insurance brokerage clears those tests. More than $2 trillion of premiums a year flow to carriers; brokers take a vig; the industry has minted dozens of large firms and has been brutally hard for startups to enter. The reason is the value chain. Carriers make money two ways, underwriting and investing, and historically they have made almost nothing underwriting. Profit comes from investing float, so the game is gathering high-quality gross written premium — which makes the broker who controls that flow unusually powerful.

Another feature: the customer does not pay the broker. The carrier does. Retention sits around 90%, the business is relationship-centric, and brokers do not really compete on price. That is a large market, fully armed incumbents, and almost no startup wedge — Sequence’s template. Baldwin added two extras. CEO Trevor Baldwin had already pushed Anthropic end-to-end through the firm, and the company runs a single instance of Applied Epic, the agency management system that functions as a broker’s core operating system. Anyone who has done enterprise IT knows how much intent those two facts imply. Take-privates also need certainty of capital. Most institutions cannot write a multi-billion-dollar equity commitment letter, which is how Sequence ended up with Michael Dell and Dan Batar, head of global direct investing at the Dell Family Office, as co-controlling partners after they underwrote both the deal and the technology motion.

What generalizes is Atlas, a platform Sequence built at the bank and expects to carry across industries. Engineers from Scale AI, Palantir, and similar shops had noticed that if you break a business into atomic units, roughly 80% is homogeneous and 20% is vertical. Atlas has four layers. A data ontology defines the organization in code so models can tell that a customer in one system is the same person in another, and how that person ties to claims or loans. An agent builder sits on that ground truth. Lattice orchestrates workflows with those agents. Artifacts is the application builder on top, so operating-company engineers can build on the stack. Core infrastructure from the bank, Lee says, is reusable at Baldwin and at whatever comes next.

He still calls the harder problem human engineering. Change draws anxiety: what does this mean for us? Sequence’s playbook is to elevate the job, strip out rote work, let people do the human parts, and make employees feel the organization is winning. The engineers Sequence hires split into two personas. Forward-deployed people from Scale and Palantir want their output aligned with the value they create, and they want to drive change a vendor could not. Application-software engineers hear something venture capitalists like less: Sequence’s model is largely immune to model progress. Better models are more tools, not an existential threat.

Large buyout shops all talk about AI transformation. Lee used to work at Apollo and still has friends across the industry; he thinks those firms will keep doing well. The genetics, he argues, are still different. A fund is paid to deploy capital: find a great asset, price it, set the capital structure, create enough value, and satisfy LPs. A typical fund investment, however long the pitch deck says the horizon is, is usually about packaging the company to sell in three years. That is a different frame for how much technology transformation you are willing to fund. Sequence’s job is to find world-class organizations run by exceptional people and take a very long view of forging the market leader. Lee hopes to hold Sequence equity for the rest of his life.

Talent is the other constraint. A typical PE firm celebrates the investor. Lee’s engineers want to be in the room when decisions are made, including whether a business is good or bad. He doubts a 25-year-old engineer has that voice at a large-cap buyout firm. Sequence also looks for management teams who are already great on their own field — he has no expertise running a broker or a bank — and who have already laid groundwork: in the cloud, centralized data, OpenAI or Anthropic in daily use at a time when ROI was still noisy and employees were anxious. Baldwin is the example. Trevor Baldwin drove those changes before a buyer showed up.

Starting the company, he says, multiplied his empathy for founders overnight. Taking an idea from nothing to something, attracting people and capital, is not easy; some days are exceptionally lonely. It is also the most fun he has had, and everything Elon Musk and Jensen Huang say about the pain is true. The Dell partnership was born of a practical need: a take-private requires an equity commitment letter most people cannot write. After shopping the Baldwin thesis with several partners, Sequence was introduced to the family office, which underwrote the deal and the transformation plan and will co-control Baldwin for a long time.

Asked how a career across publics, privates, and now operating colors the work, Lee describes an amalgamation. Apollo taught him deal structure, credit agreements, and management incentives. Lone Pine taught him to study NVIDIA, Microsoft, and Visa, and to internalize that the greatest companies compound faster than anyone thinks. Building Lone Pine’s private business put him at the technology frontier, watching founders will something from nothing. If he ever went back to private investing — he will not — the lesson would sound like early-stage venture: ideas are cheap, execution is hard, and truly exceptional people figure it out. Back them in huge markets even when it looks crazy. He met Huang in 2017, when NVIDIA was not yet what it is, and saw consistency, clarity, loyalty around him, and a habit of re-founding the company as the market moved. That, he says, is the name of the game in both private and public markets.