Fireworks AI Valuation 2026: $17.5B Series D and the $1B ARR Inference Business

Fireworks AI is now worth $17.5 billion after a $1.505 billion Series D closed on July 16, 2026 — roughly 4x what it was worth seven months earlier. That's the short answer. The longer answer is that this jump is backed by real revenue: ARR crossed $1 billion, up about 5x year-over-year, as the company's daily token volume nearly tripled.

I've watched inference-layer valuations move faster than almost anything else in AI infrastructure this year, and Fireworks is the clearest data point yet. This is a look at the round, who's backing it, how the revenue actually scaled, and how Fireworks stacks up against Together AI and Baseten — the two companies it's most directly fighting for enterprise inference workloads.

What Is Fireworks AI's Valuation in 2026?

Fireworks AI's valuation is $17.5 billion as of its $1.505 billion Series D, announced July 16, 2026 and led by Atreides Management, Index Ventures, and TCV. The company had reportedly been in the market seeking around a $15 billion valuation just weeks earlier — the deal priced above even that target once investor demand became clear, which is itself a sign of how competitive the AI inference infrastructure category has gotten.

Seven months prior, Fireworks was valued at roughly $4 billion. A jump of that magnitude in under a year is unusual even by 2026 AI-funding standards, and it tracks almost exactly with the company's revenue growth over the same period — this isn't a valuation running ahead of fundamentals, it's one keeping pace with them.

Fireworks AI's Revenue Growth Through 2026

The revenue trajectory is what makes the round defensible rather than speculative. Fireworks ARR sat around $315 million in February 2026, up 416% year-over-year at that point. By May 2026 it had climbed to roughly $800 million. By the July Series D announcement, ARR had crossed $1 billion — meaning the company added several hundred million dollars of run-rate revenue in roughly ten weeks.

Figures blended from Fireworks AI's Series D announcement, Sacra revenue tracking, and CNBC reporting through July 2026.

Who Invested in Fireworks AI's Series D?

The $1.505 billion round was led by Atreides Management, Index Ventures, and TCV, with a long list of participants: Nvidia, Lightspeed Venture Partners, Bessemer Venture Partners, Menlo Ventures, Insight Partners, Lone Pine Capital, Evantic Capital, Operator Collective, Ontario Teachers' Pension Plan, Original Capital, Prysm Capital, Quantum Capital, 20VC, and TIME Ventures.

Nvidia's participation stands out. Fireworks runs its inference engine primarily on Nvidia GPUs, so this is another case — alongside Nvidia's stakes in CoreWeave, Together AI, and other AI infrastructure players — of the chipmaker investing directly in the companies renting its own hardware back to enterprises. It's a strategy that both secures demand visibility and captures upside from the inference layer, not just the chip sale.

Fireworks AI's Full Funding History

Fireworks has now raised more than $2.1 billion since its founding by former Meta and Google engineers who worked on PyTorch and large-scale ML infrastructure. The company's Series C closed at $250 million in late 2025, valuing it around $4 billion — a number that looks almost quaint against the $17.5 billion Series D just seven months later. Total funding history includes seed and Series A/B rounds from Benchmark and Sequoia before the company scaled into the multi-hundred-million-dollar territory typical of 2026 AI infrastructure rounds.

What's notable is the size jump between rounds: the Series D at $1.505 billion is roughly 6x the dollar size of the Series C, while the time between rounds was under a year. That compression — larger checks, shorter intervals — has become the norm for any AI infrastructure company that can show real, compounding usage growth rather than just a roadmap, and Fireworks' token-serving numbers gave investors exactly that kind of evidence.

Who Uses Fireworks AI and Why It Matters for the Valuation

Fireworks' customer base skews toward enterprises running production AI features at scale rather than experimentation — companies that need consistent low-latency inference across hundreds of millions of daily requests, not a handful of API calls during a pilot. The company has publicly cited partnerships tied to Microsoft and Nvidia's cloud and hardware ecosystems, positioning itself as the layer that turns open-weight models like Llama, DeepSeek, and Qwen into production-grade infrastructure without customers having to manage GPU fleets themselves.

That production-first customer mix is part of why the $17.5 billion price tag has support: usage-based inference revenue from live enterprise traffic is a fundamentally more durable revenue base than committed-but-unused cloud credits, which have inflated some other AI infrastructure valuations in 2026. Fireworks' 40 trillion tokens served per day is a real operational metric, not a projection, and it's growing roughly in line with the ARR curve rather than diverging from it.

Fireworks AI vs Together AI vs Baseten: Inference Platform Valuations Compared

Fireworks isn't alone in this space. Together AI raised an $800 million Series C on July 1, 2026 at an $8.3 billion valuation, with ARR reportedly around $1.15 billion in annual bookings. Baseten, a smaller but fast-growing competitor, is valued around $1.3 billion with a fraction of Fireworks' revenue scale. All three are chasing the same core opportunity: enterprises that want fast, cost-efficient inference on open-weight and fine-tuned models without building their own GPU fleet.

For more on Baseten's own funding trajectory, see our Baseten valuation breakdown, and for how the broader market is pricing AI infrastructure companies, check our AI Valuations dashboard.

How Fireworks AI Actually Makes Money

Fireworks operates a usage-based B2B inference platform — customers pay per token and per GPU-hour to run hundreds of open-source and fine-tuned models across text, image, audio, and multimodal formats. The platform's pitch is speed and cost: Fireworks markets itself as the fastest inference engine for open-model deployment, which matters directly to enterprise gross margins once a company is serving billions of tokens a month.

Beyond raw inference, Fireworks monetizes fine-tuning, reinforcement learning tooling, and model evaluation — layers that increase the switching cost once a customer builds a production pipeline on top of the platform. That stickiness, more than any single feature, is what's letting Fireworks post ARR growth this steep without a corresponding blowout in customer acquisition cost, according to the company's own funding disclosures.

Fireworks AI's Revenue Multiple: Is $17.5B Reasonable?

At $17.5 billion on $1 billion-plus ARR, Fireworks priced its Series D at roughly 17.5x revenue. That's rich by traditional SaaS standards — public infrastructure software companies typically trade at 8-12x forward revenue — but it's in line with, or even conservative relative to, other 2026 AI infrastructure comparables. Together AI's $8.3 billion valuation against roughly $1.15 billion in annual bookings works out closer to 7x, while several foundation model labs are pricing well above 20x on revenue that's growing faster but from a much smaller and less durable base.

The multiple gap between Fireworks and Together AI is worth watching, since both companies compete for nearly identical enterprise inference budgets. If Fireworks' growth rate and margin profile hold up through the next two quarters, the higher multiple looks earned; if Together AI's bookings growth overtakes Fireworks' ARR growth, the market will likely compress the gap fast, given how quickly capital rotates between the top two or three players in a category this contested.

What This Means for AI Infrastructure Investors

A $17.5 billion valuation for a company that was worth roughly $4 billion seven months ago tells you two things about the state of AI infrastructure investing in mid-2026. First, the inference layer — not just model training — is now where a meaningful share of enterprise AI spend is landing, and investors are pricing that shift aggressively. Second, revenue quality is still the gating factor: Fireworks' round tracks a real 5x ARR increase, not a valuation markup detached from fundamentals, which is the distinction that separates this deal from some of the frothier pre-revenue AI raises in our AI valuation multiples framework.

The risk to watch is concentration and competition. Together AI, Baseten, and the hyperscalers' own inference offerings (Google Vertex, AWS Bedrock, Azure AI Foundry) are all chasing the same enterprise budget, and token pricing is a race to the bottom over time. Fireworks' current edge — speed and the ability to serve 40 trillion tokens a day — has to keep compounding faster than the category commoditizes, or a $17.5B multiple on $1B ARR (17.5x) starts to look stretched relative to where SaaS and infrastructure multiples typically settle.

The Bottom Line

Fireworks AI's $17.5 billion valuation is backed by a revenue curve that's genuinely rare in AI infrastructure right now: ARR that roughly tripled between May and July alone, on top of an already-fast 2026. That's a different story from many pre-revenue foundation model raises, and it's why Atreides, Index, TCV, and Nvidia priced the round above the company's own initial $15B target.

The number to watch next is gross margin, not top-line ARR — inference is a compute-intensive business, and whether Fireworks can keep growing revenue 5x year-over-year while defending margin against Together AI, Baseten, and the hyperscalers' native inference products will determine whether $17.5B looks conservative or aggressive a year from now.