a16z · Nov 10, 2026 · 59m · 50m saved
With Dharmesh Shah, Co-Founder and CTO of HubSpot
The US set a record for new company formations in 2025, at around 5.6 million, and over 70% of new companies are now expected to stay one-person businesses. Dharmesh Shah, co-founder and CTO of HubSpot, told Joe Schmidt, a partner at a16z, on the a16z podcast that this is the same kind of dismissed market small-business software was 20 years ago.
Much of the software industry is worried that the frontier AI labs will absorb the applications built on top of them. Dharmesh argues the opposite: the labs' incentives point elsewhere, and he has rarely enjoyed building more.
"I've been building commercial software for 30-plus years now, literally every day, and this is the most fun I've ever had in terms of building things. What a time to be a builder." — Dharmesh Shah
Dharmesh built HubSpot from an MIT business-plan idea into a 20-year-old company, and he is also an investor in OpenAI. He is now running HubSpot's in-house startup aimed at one-person companies.
The full interview is covered here so you can skip it. 59 minutes of audio, 9 minutes of reading.
Here are the 7 insights that matter.
Key Takeaways
An idea that only works because "ChatGPT can't do that yet" is fragile
The safer test is whether a frontier lab has a structural reason to stay out
HubSpot started as ERP software for law firms and pivoted when lawyers said they wanted clients instead
60% of paying HubSpot marketing customers used no CRM at all, so HubSpot built a free, simple one
AI chat apps are the new browser, not the new application, in Dharmesh's view
Never use inference where a database query would do
A 30-day-old solo CRM is the best monetization test Dharmesh has run at HubSpot
Over 70% of new companies are expected to stay one-person businesses
1. Stay Off the Labs' Road
The conversation opened on an essay by Joe arguing that software founders can build away from the frontier labs.
"There's plenty of room to build software outside the path the Frontier AI labs are racing down." — Joe Schmidt
Dharmesh said the labs are acting rationally. They have spent hundreds of billions of dollars on pre-training and now have to earn it back, either by selling tokens directly to consumers and Fortune 500 buyers or by letting software companies resell inference.
Dharmesh Shah says an idea that depends on "yet" is fragile
Founders get into trouble, he said, when an idea works only "because ChatGPT can't do that yet." The safer test is whether a frontier lab has a structural or incentive reason not to enter the market. He used OpenAI and HubSpot as the example.
"I'll pick on Sam Altman since I'm an investor in OpenAI and know Sam." — Dharmesh Shah
Despite knowing Altman, Dharmesh thinks it unlikely that OpenAI would try to unseat HubSpot in CRM, because its ambition is aimed elsewhere.
