đ Iâm Ivan. I study how top 1% startups grow.
Hello there!
This week Iâm diving into Fireworks AI, a company running open AI models for other companies (i.e. Cursor, Uber, Notion) that recently raised $1.5B at a $17.5B valuation (4 years after its founding, another AI-wave rocket-ship):
In a nutshell:
Product: think about it like an âengine roomâ for AI. Companies that donât want to depend on OpenAI / Anthropic take a free open model (i.e DeepSeek or Llama) and Fireworks runs it for them (fast, cheap and inside their product). Today it apparently already handles 40T+ tokens a day (tokens are the chunks of words AI reads + writes) for 10K+ companies.
Money: the model is whatâs quickly becoming a classic = pay per use (i.e. $0.30 per million tokens in, $1.20 per million out on a model like DeepSeek), renting chips by the hour (i.e. $8/hr for an Nvidia H100), or pay to train your own custom model. Theyâve already passed $1B in annualized revenue as of July this year with around ~200 people ($5M per employee!).
Driver: loved how they frame it here: âCompanies are no longer renting general intelligence. Theyâre building their own.â
8 growth levers in this drop:
Sold Metaâs internal speed tricks to everyone
Made switching painless at the moment costs typically start to hurt
Had every hot new free model running on launch day
Built custom for the fastest-growing customer + then sold it to everyone
Made engineers part of the âunofficialâ sales team (inside the customerâs Slack)
Made a custom model cost the same as a generic one
Turned big clouds into both suppliers and a sales channel
Bet on deciding which model runs for each task
đ Quick note on editorial + methodology: this analysis focuses on the 80/20 mechanics that explain their growth (itâs not a comprehensive profile, not an endorsement or investment advice). I use AI like a fund leverages an analyst for groundwork, the direction + judgement are mine. Company-reported figures are marked as such, treat directional estimates as directional.
Zero to one
Who: 7 engineers who used to run the AI plumbing at my good old employer Meta + Google (6 from Meta and four of them actually from the PyTorch team aka the free toolkit most of the world uses to build AI models today).
Lin Qiao (CEO): led the PyTorch there growing the team from 5 to ~300 people, before that IBM and LinkedIn. She started Fireworks at age 48 (after shelving a 2015 startup plan to learn how to lead people at Facebook). She meant to stay âone year or twoâ and stayed 7.
Dmytro Dzhulgakov (CTO): from Ukraine whoâs a PyTorch core maintainer after 11 years at Meta, before intern at Google.
Benny Chen: a new Zealand-born engineer who ran ads infra also at Meta.
Plus Dmytro Ivchenko, James Reed, Pawel Garbacki (all ex-Meta) and Chenyu Zhao (ex-Google).
Where they started (fall 2022): they all apparently left Meta about 2 months before ChatGPT came out and the first plan was a platform to help companies use PyTorch. Benny, in 2024 said: âExactly how, we didnât really figure out completely... ChatGPT wasnât out yet, so we had to pivot somewhere in the middle.â They raised $25M from Benchmark and Sequoia before having let alone launching a product. Apparently, fun detail, they started in a building literally called the âSequoia buildingâ before Sequoia invested.
The wedge: PyTorch was of course free and open so companies like Walmart, Disney, Tesla, Netflix etc used it too and they kept coming back to Linâs team asking âcan you build this training platform for us? Can you build this serving platform for us?â Lin took it as a signal the industry was ready / market timing was right, âand then more importantly, we have the key.â
The MVP: a platform that launched in August 2023 (11 months in). Run and customize free models through a simple web connection with a free tier for devs:
First users: âa lot of ex-coworkers from Meta,â says Benny plus teams who trusted the PyTorch name (this is a classic Silicon Valley dense network kindling effect), and by mid-2024 the list already included Cursor, DoorDash and Quora.
Growth Mechanics
Lever 1: Sold Metaâs internal speed tricks to everyone
âOur biggest differentiation is... Fireworks off the shelf is faster than both of the offerings and second is weâre building a system, not just a library.â (Lin, Sequoia)
What happened: at Meta theyâd built these tricks that let AI run at planet scale with things like custom code that squeezes more work out of chips, compression that makes models smaller without losing (much?) quality, etc. At Fireworks they then packaged those tricks + rented them to companies through a simple connection.
The details:
In Jan 2024 they said their engine ran 4x faster than open-source alternatives
Speed matters here, youâve probably experienced it using any of the models consumer facing interfaces: âlow latency is a critical part of product viability... people are not patient enough to wait for half a minute.â
By Series B in July 2024 they were serving already 140 billion tokens / day.
Meta itself became one of their customers this year.
But this edge doesnât last on its own as we see on Artificial Analysis (independent speed rankings) Fireworks sits 6th of 11 on DeepSeekâs newest big model.
So what: turning a big companyâs internal tools into a product is a classic founding play and in this case it got them in the door / great wedge. Speed was an advantage but it isnât necessarily a lasting moat (rivals caught up within 2 years, its also a little silly to talk about moats in this market considering how fast everything moves, but worth keeping an eye out). Interesting what theyâve done to retain customers (levers 4 to 6 later).
Lever 2: Made switching easy when costs start to hurt
âThey cannot open up the floodgate because theyâre going to scale into bankruptcy.â (Lin, theCUBE)
What happened: companies often build the first version of their AI products on OpenAI, Anthropic or similar because itâs the fastest way to test an idea. If the product takes-off usage can explode and so does the bill and that can be a problem. So Fireworks built for that moment or rather with that in mind. So they made it easy to move over with a few lines of code and the open models it runs cost much less to use.
The details:
Going bankrupt on AI bills comes up in 17 of the 44 interviews we went through for this deep-dive, itâs essentially their sales pitch.
It now reaches big companies too, the founder Lin recently said in a Sequoia interview: âtheir CFO is blocking their AI feature launch because of the costâ
The pain in the market has become so big / obvious that apparently they barely marketed. Lin recently said on 20VC âwe feel product speaks for itself... we didnât spend much time marketing at all.â
They apparently priced for the comparison table, moving in March 2024 to a flat price per token âbecause people would often make comparisons using output token costâ.
And just to show you how fast these switches are going for out there apparently Innovative Solutions (AWS partner) moved â90% of Anthropic inference spendâ to Fireworks in 2 weeks (case study, May 2026).
So what: they never fought OpenAI for the prototype. They waited for the moment the bill hurts and made leaving nearly free. In strategy terms theyâre the cheaper substitute for the closed labs, and a substitute only wins if switching is painless.
Lever 3: Had every new âhotâ free model running immediately on launch day
âWe are very proud of day zero launch always. Like we kind of become famous on day zero launch.â (Lin, Startup Grind Q&A, May 2026)
What happened: every time a new free model came out like DeepSeek, Kimi, Llama etc. etc., they tried to have it running for customers the same day.
The details:
DeepSeek R1 came out in Jan last year and they had a full breakdown 4 days later.
OpenRouterâs CEO says Fireworks âcaptured most of the inferenceâ on DeepSeek in the early days at least âas far as our metrics told usâ (Sequoia)
Two months after R1 they matched DeepSeekâs API price at $0.55 in and $2.19 out per million tokens.
They also know when to break their own rules intelligently for example with DeepSeekâs V4 release since it had bugs, so they held it back about 3 days.
They now host 200+ models and add one pretty much every week these days.
So what: the open-model boom to certain extent was luck in terms of a tailwind, but what they did differently was treat each launch as a free acquisition moment and try to never miss one (turning othersâ announcements into their own marketing).
Lever 4: Built custom for the fastest-growing customer, then sold it to everyone
âWe have finite engineers like everybody else. We would like prefer to have engineers make training more efficient... rather than like spin up like a inference effort.â (Federico Cassano, Cursor, Sequoia 2026)












