General Catalyst targets the $16tn services market by buying traditional firms and automating 30-70% of tasks to lift margins from 5-10% to 30-40%.
It has about eight roll-up bets and five incubations, including Crescendo in call centers and Long Lake in HOA management.
Marc Bhargava says the model avoids debt and cost cuts, keeping management and growing revenue.
General Catalyst bets AI roll-ups can lift $16tn services market
General Catalyst is betting that AI turns services from a bad venture business into a good one. Services generate about $16tn of annual revenue but run at 5-10% margins, too thin for venture capital, and automating 30-70% of tasks could lift them to 30-40%. The firm's vehicle is a compounder that begins as vertical software and then buys the fragmented service businesses it sells to.
Takeaways
Services is a $16tn pool at thin margins. General Catalyst's thesis is that AI can lift services margins from 5-10% to 30-40% by automating 30-70% of tasks, a market roughly 16 times the size of software.
Buy distribution, automate it, buy more. General Catalyst's model starts as vertical software that reaches product-market fit, then acquires traditional service firms, improves them to generate free cash flow and uses that cash to buy more, following Danaher, TransDigm, Constellation Software and Berkshire Hathaway.
Hybrid teams pair AI leaders with operators. The firm studied 70 service categories and picked 10 verticals where 30% or more of tasks look automatable, pairing applied AI leaders from Rippling, Ramp, Scale AI and Figma with M&A veterans.
Eight roll-up bets, five incubated in-house. Portfolio companies include Crescendo in call centers, Long Lake in HOA management, Udia in legal services, Hippocratic in nursing and Dwelling in London property management. Automation runs 80% in bookkeeping. Call centers reach 50-70% and MSP/IT 38%.
Sellers roll 30% equity for another 10-15x. Pilot clients often cannot implement automation themselves and choose to sell, taking cash and equity. Deals carry no debt, keep the same management and pursue revenue growth rather than margin extraction, aiming for a 20% cost increase to double revenue in three years.
Speakers split on how many firms can join. Labenz guesses only about 20% of local service firms have the leadership and culture to be attractive targets, and that 90%-plus task automation could displace many accounting workers within two years. Bhargava says the tools are intuitive and a majority can participate.
What resists automation
Physical trades such as plumbing, roofing, HVAC and firefighting stay human.
High-churn industries such as advertising and PR agencies are poor fits.
Work where current models perform worse than humans stays out of reach.
Bhargava still expects hybrid human-AI workforces in services, since regulation and complexity require a human in the loop.
Predictions and open questions
Tap a name to go there · tap a line to see why they are linked
On this episode 2
Show 1
Bhargava expects near-full automation in coding, but hybrid human-AI workforces in services.
Bhargava puts the odds of all providers closing their models below 3%, given seven or eight strong labs including xAI, DeepSeek, Alibaba and Mistral.
The Magnificent Seven spent $150bn on AI research and capex last fiscal year, heading toward $250bn this year.
Model progress reaccelerated after a flat stretch.
Labenz expects high consumer surplus, with better service at similar or lower prices.