Fireworks AI: Putting gen-AI in the reach of any developer

In partnership with

(3 minute read)

• This AI startup saves AI businesses 4x their money

• They just raised $52M at a $552M valuation led by Sequoia Capital

• They already have partnerships with tech giants like Nvidia, AMD and AWS

So what’s the startup and who are the founders behind it? Here’s the story of Fireworks AI 📈:

💥 Fireworks AI was founded in 2022 by Lin Qiao, Dmytro Dzhulgakov, Chenyu Zhao and Dmytro Ivchenko with the vision to provide developers worldwide with gen-AI models to build and innovate on top of.

🦾 Fireworks AI is the fastest and most efficient inference engine to build production-ready AI systems.

They offer a generative AI platform as a service and a platform for fine-tuning models. These 2 go hand-in-hand to deliver customized models using proprietary data towards a specific use case.

In simpler terms, instead of larger, more generic models like ChatGPT and Llama 3, Fireworks AI provides smaller, production-grade models which can be customized for a client's specific needs.

This results in the best, lowest latency for gen-AI applications for the lowest total cost ownership.

The AI startup has 3 major values:

• Quality

• Latency

• Total Cost Ownership

This keeps them heavily focused on a business’ production, helping enterprises and developers launch their products at massive scale.

How’d they start out?🐣

Before Fireworks AI, all the co-founders worked as engineers at Meta for 7-10 years each. At Meta, the co-founders worked on PyTorch, one of the biggest ML libraries for applications like computer vision and NLP.

During their last 5 years at Meta, their team had been building AI infrastructure to help Meta transition to an AI-first company 🤖 .

To do this, the engineers had worked on converting PyTorch from a research only library to supporting all types of production workloads and run them at scale…

By the time they left Meta, they were running 5 TRILLION inferences per day across all data centers globally.

Looking at the industry, it was clear a transition to AI was happening.

The industry however, was also going through excruciating pain because the talent to support this AI-first transition didn't exist. They didn't have GPU’s to run AI models, or the knowhow of the best software tools to run on top of the GPUs. 

This meant extremely high latencies and high costs…

This is where Fireworks AI comes in💥. Fireworks AI is on a mission to support the whole industry to go through with this AI-first transition.

At Meta, it took them 5 years to run PyTorch at scale, but the co-founders knew this time could be compressed by several orders of magnitude.

They wanted to compress it from 5 years to 5 weeks or even 5 days to enable enterprises and developers to build innovative and disruptive solutions built on top of gen-AI models 🤝.

The way it’s able to do all this is because they limit their model size to 7-13 billion parameters. In comparison, GPT-4 uses 1 trillion parameters.

Fireworks AI heavily focused on the production process, and through using much smaller, more specialized models, started to take off 📈.

The startup allows businesses to entirely focus on building and innovating their product, while running the business’ operations autonomously.

Stats 📊

The startup now serves over 100 models in text, image, audio, and embedding, optimizing for latency and throughput.

📈 Their chat product costs 40x less than other chatbots like GPT-4 and Llama-3 and has a 15x higher throughput.

Fireworks AI boasts the largest open source model API with over 12,000 users, while generating 140B+ tokens and 1M+ images per day.

💰 They already have partnerships with Nvidia, AMD, AWS, Google Cloud and Oracle Cloud for model infrastructure optimization.

👨‍💻 They’re also already being used by developers at tech giants like Doordash, Quora and Upwork for their specialized models.

💸 On July 8 2024, Fireworks AI raised a $52M Series B at a $552M valuation led by Sequoia Capital, bringing their total raised to $77M.

Notable investors in previous rounds include Benchmark, Databricks Ventures, Scale AI CEO Alexandr Wang, and former Meta COO Sheryl Sandberg.

Before you go 👋 

Your Brilliant Business Idea Just Got a New Best Friend

Got a business idea? Any idea? We're not picky. Big, small, "I thought of this in the shower" type stuff–we want it all. Whether you're dreaming of building an empire or just figuring out how to stop shuffling spreadsheets, we're here for it.

Our AI Ideas Generator asks you 3 questions and emails you a custom-built report of AI-powered solutions unique to your business.

Imagine having a hyper-intelligent, never-sleeps, doesn't-need-coffee AI solutions machine at your beck and call. That's our AI Ideas Generator. It takes your business conundrum, shakes it up with some LLM magic and–voila!--emails you a bespoke report of AI-powered solutions.

Outsmart, Outpace, Outdo: Whether you're aiming to leapfrog the competition or just be best-in-class in your industry, our custom AI solutions have you covered.

Ready to turn your business into the talk of the town (or at least the water cooler)? Let's get cracking! (And yes, it’s free!)

Thank you for reading.

Remember, if you know anyone who might like this newsletter, sharing is caring.