MA billion-dollar valuation for a company that builds computers for AI agents. That's the lead today.
CDaytona. Sixty million in a Series B, led by FirstMark Growth.
MAnd the pitch is simple: every AI agent needs a computer. So they're building millions of them.
CI actually love this pitch. It's so basic it's almost boring.
MThat's what makes it interesting. Everyone's fighting over models. This is the plumbing.
CRight. An agent doesn't just answer questions. It runs code. It needs a sandbox, an operating system, a GPU sometimes.
MAnd you don't want some AI model executing arbitrary code on your laptop. You want it quarantined.
CSo Daytona spins up sandboxes: Linux, Windows, Mac. Secure environments where the agent can do its thing.
MThe founder, Ivan Burazin, says it's the Agent Cloud. A computer for every agent.
CThat's the bet. That agents become real workers, not chatbots.
MI think that's the moment we're in. Last year it was 'AI can write.' This year it's 'AI can do.'
CAnd every 'do' needs a machine to run on. That's why this stuff gets funded.
MNinety-one million total now. They raised the sixty at a billion-dollar valuation. So someone's convinced.
CAngels too. TechCrunch says a who's-who of angels came in. I want to know which ones.
MOf course you do. You probably have their podcasts queued.
CGuilty. But look, here's the risk. What if agents don't take off? All this infrastructure sits idle.
MThat's every venture bet. You're paying to be the rails before the trains show up.
CThe difference is these trains are improving every month. Code generation is getting scary good.
MSpeaking of scary good code: Greptile just raised fifty million at a seven-hundred-fifty million valuation.
CMenlo Ventures led. They're doing AI code review and testing.
MHere's the number that jumped out at me. They review more than one percent of all pull requests on GitHub. Daily.
CThat's... enormous.
MNvidia, Coinbase, Vanta are customers. So real teams are shipping with this thing looking over every change.
CAnd I love this because code review is where the arguments happen. Humans get tired, miss things, play politics.
MNow you're just describing your old engineering team.
CFair. But if an AI is checking every diff, you catch bugs before they ship. That's huge.
MDoes it actually write the tests too, or just nitpick?
CIt reviews, it tests. It's the boring part of software engineering that nobody wants to do.
MI'll say this: a billion dollars for AI sandboxes, three-quarters of a billion for AI code review. The money's moved downstream.
CDownstream is where the money gets real. Models are becoming commodities. The work is in the doing.
MAlright, here's one that might start a fight. There's a site called Local Wikipedia.
COh, I saw this on Hacker News. It shows Wikipedia entries within a half hour walk of you.
MSo instead of reading about the Roman Empire, you read about the abandoned mill down the street.
CAnd people are weirdly into it. One guy found the nearest entry was a three-hour walk away and got annoyed.
MThree hours? That's not a walk, that's a hike.
CBut then he listed all these un-tagged things near him. A canyon, a log. He wanted them in the wiki.
MThere's something human about that. You discover your own neighborhood has history worth recording.
CAnother person said nearby articles feel more important. A decades-old plane crash two miles away hit different.
MThat's just... proximity bias. But I get it. Local history feels personal.
CYou know what? This is the antidote to all the AI infrastructure talk. Small, human, curiosity-driven.
MAnd free. Built by someone who just wanted to know what's around. That's the whole thing.
CI bet there's a listener right now checking what Wikipedia says about their block.
MDifferent world entirely from the billion-dollar rounds. But same impulse: understand the machine, map the territory.
CAnd there's one more funding story I should flag, but I'm not sold on it yet.
MGo.
CTown. Supposedly raised a hundred and eight million at a billion-dollar valuation. Four months after a fifty-five million Series A.
MThat's a wild pace. What is Town?
CAI assistant for teams, from a former Plaid CTO. Claims it hit seven-figure annual recurring revenue in weeks.
MIn weeks? That's the part I'd want to see the receipts on.
CAgreed. The account who posted it, Traded VC, says Forerunner and Menlo co-led. But there's no official confirmation yet.
MRumor mill says a lot of things after a big week of AI funding. Double the check before you double the valuation.
CFair. But if it's true, that's two fresh AI billion-dollar companies in one week. The fuse is lit.
MAnd the fuse leads straight to the question we keep asking every show: is this real value, or just money chasing heat?
CThe rubber meets the road when these agents actually ship. Sandboxes, code review, team assistants. Somebody has to work.
MProbably several somebodies. And they'll be running on infra like Daytona's.
CThen Local Wikipedia reminds you: half an hour from your door, there's a story the big models never learned.