The AI Rollup Wave Transforming Main Street

0:00 So how is AI going to actually change the economy? It may not be with flashy moonshots like robotics. This is Woof AI. New devices. Rabbit R1, a dedicated AI device. Or even data centers in space. It's a no brainer for building solar powered AI data centers in space. Instead, it might just be the boring companies that are ripe for transformation and Silicon Valley's newest obsession. We think every company in America needs an AI transformation strategy. The same VC firms that wrote checks to OpenAI and Anthropic, they're now buying up property managers, accounting firms, even insurance companies.

0:37 It's a bet that you don't transform businesses by just layering AI on top of them, but by rebuilding them from the inside out. I'm Deirdre Bosa. Silicon Valley is coming for the real economy. It's called the AI rollup. General Catalyst, Thrive Capital, Lightspeed, Andreessen Horowitz, they're all running versions of it.

1:08 And the playbook is simple, create a holding company, pour in capital, buy up Main Street businesses that tech has traditionally ignored, like Amex GBT, a corporate travel platform, or the asset manager, Janus Henderson, then rebuild them around AI. So a labor heavy business can start to scale more like software. General Catalyst has created roughly a dozen of these holding companies in the last three years. Managing director Madhu Namburi has a name for it. Software, SAAS was the definition of software as a service. I believe it's going to be service as software, and I think the era has come, and I think it's going to happen a lot more.

1:45 The point isn't just to make these companies a little more digital, it's to change their economics. Software businesses, they became some of the most profitable companies in history because they could scale without adding costs at the same rate. Think Microsoft, Oracle, Adobe, they build it once they sell it endlessly and the margins, they run 70, 80, even 90% historically. Service businesses, they never had that luxury. They're built on human time, like answering emails, processing forms, scheduling appointments.

2:13 More clients meant more people, and that meant more cost. AI is the attempt to break that link, give the same team more leverage. Services is a $20 trillion market. One of the biggest issues in services was its inability to scale. Growth is slower. Margins are lower because every time these services companies had to grow, they had to add a lot of bodies. AI is enabling a fundamental transformation to grow these companies without adding more headcount.

2:45 Put AI inside a services business, and the bet is that its economics start to look more like software. The clearest example is Long Lake. A three year old holding company backed by General Catalyst and Alpha Wave, it has acquired over 30 businesses spanning HOA management, construction, corporate travel. The key is Long Lake's own AI platform called Nexus, built for each industry's workloads. CEO Alex Taubman he says that it outperforms models like ChatGPT and Claude for those businesses.

3:14 HOA management, for example, is a is a relatively niche. It's actually a big business. And so we sort of take the best in class models. We harness them in our platform, we give them specialized workflows and skills and tools that are heavily customized for these end markets. And it drives much better performance in the field. So what we're actually seeing, our Nexus platform is actually performing five times better than the off the shelf models. And that is the core of the bet.

3:37 ChatGPT and Claude, they may be impressive in the abstract, but the money is an automating workflows, HOA disputes, invoices, travel changes, customer emails, the dull stuff that eats hours and makes services hard to scale. Another piece of the playbook, embed engineers inside of those businesses. Most of Long Lakes engineers, they come from Ramp and Palantir companies known for putting engineers close to the customer. Long Lake says it plans to hold these companies permanently, more like a Berkshire Hathaway than a traditional VC or PE fund.

4:08 We plan to own and operate these businesses forever, so we're not that focused on value up front. We think about where can we drive the most value for customers, where can we drive the most growth? And then for us, that's where we lean in. And one of our ambitions is we'd like to try to be one of the premier or the best buyer for really any services company that fits our thesis. On the company, the implementation, the productivity gains.

4:34 If the AI roll up sounds familiar, that's because part of it is. Private equity has been doing roll ups for decades. Buy companies, consolidate them, cut costs, sell them later. But in the early 2020s, some of the biggest private equity firms made a different bet. They bought enterprise software at peak prices. Vista bought Citrix, Thoma Bravo bought Anaplan and Coupa. Silverlake bought Qualtrics. The thesis was simple, recurring software revenue was the safest cash flow in business.

5:03 But then AI arrived, and suddenly that safe software revenue looked a lot less safe. And now some of those firms are trying to retrofit AI into their portfolios through partnerships with OpenAI and anthropic. But that can look like a consultant's fix. Someone else's AI dropped into someone else's company by investors who don't really own the technology or the workflow. The AI roll up is a different playbook. Don't just buy the software company's AI might disrupt. Buy the services companies, AI might transform.

5:36 But this is still a risky bet. Venture capital is built for outliers and moonshots. One company that can return the entire fund, Main Street businesses, they don't usually work that way. A good services company might be a great business, but it won't necessarily return ten times the investment. So the returns may look more like private equity than venture capital. But Namburi argues General Catalyst has an edge. We believe we have an unfair advantage. Our hope and target is to create that disproportionately more attractive returns.

6:08 General Catalyst was the first investor, and from that point versus what it's worth, it has been an absolute grand slam. The second risk is operating. Buying a property manager is one thing. Running it better is another. Private equity firms like Vista and Thoma Bravo, they spent decades building the operating teams to do that. Venture firms, they're trying to build that muscle now with AI as the advantage, Taubman argues the timeline is different this time. Three years in AI is actually like three decades of pre-AI.

6:38 That may be right. AI can move faster than old software, but the companies it's moving into are still real companies. They have customers, employees, regulation, legacy systems, local relationships. That's what makes this trade so interesting. Public markets, they see boring slow growth businesses. The AI roll up sees something else. Services companies whose cost structure may be about to change. If they're right, the big AI winners, they won't just be the model labs. They'll be the investors who own the businesses where AI actually changes the math.