Fal, the multimodal model hosting platform, and Fireworks AI, which specializes in open-source model inference, are each preparing funding rounds that could roughly double the valuations they commanded earlier this year. The dual raises underscore how quickly capital is flowing into AI inference infrastructure — the layer of the AI stack that runs trained models and produces outputs for end users — a corner of the market where revenue growth is separating a small group of companies from the rest of the pack.

The Numbers Behind the Raises

Fal secured $140 million in a Series D round in December 2025, led by Sequoia, at a $4.5 billion valuation. That figure already represented a tripling of the company's worth from just months earlier. By March 2026, Fal was reportedly targeting a raise of $300 million to $350 million at an estimated $8 billion valuation.

Fireworks AI has moved even faster. The company raised $250 million at a $4 billion valuation in October 2025. Discussions around a potential round at a $15 billion valuation surfaced in May 2026, and in July 2026, Fireworks closed a $1.505 billion Series D that valued it at $17.5 billion.

The soaring valuations are not fueled by investor enthusiasm alone. Both companies are posting revenue numbers that support the attention. Fal's annual recurring revenue reportedly doubled to approximately $400 million within the span of a few months. Fireworks surpassed $1 billion in ARR, a fivefold increase year-over-year, while processing trillions of tokens daily. Because inference providers typically bill by usage, that revenue serves as a near-real-time gauge of how heavily enterprises are actually deploying AI models.

Why Inference Is the New Frontier

Founded in 2021, Fal focuses on hosting multimodal models capable of handling text, images, video, and audio. Fireworks AI, launched a year later in 2022, built its platform around making open-source model inference fast and cost-effective for enterprise clients. The segment sits on the usage side of the AI economy: rather than the one-time compute of training a model, inference is the recurring work of serving responses to users, so demand scales with how widely AI features spread through everyday software.

Both companies serve major enterprise customers, with Adobe and Shopify among the organizations relying on their infrastructure — a measure of how much AI serving capacity large enterprises now source from specialist providers. Their investor rosters tell a similar story: Sequoia, a16z, Lightspeed, Index Ventures, and Nvidia have all backed the two firms. Nvidia's involvement is particularly telling, since the chipmaker has a strategic interest in ensuring robust inference infrastructure exists to drive demand for its GPUs.

A Market Pulling Away From the Pack

The implied revenue multiples illustrate the widening gap. If Fireworks reached $1 billion in ARR with 5x year-over-year growth that even partially sustains, its $17.5 billion valuation implies a 17.5x revenue multiple. Fal's situation is similarly striking: doubling ARR to around $400 million while targeting an $8 billion valuation implies a roughly 20x revenue multiple. Multiples at that level embed expectations of continued rapid growth, which is why the revenue lines rather than the headline valuations are the figures to track from here — whether Fal's reported round closes near the valuation under discussion, and whether ARR growth at both companies keeps pace as their bases grow larger.

Together, the figures point to AI inference as one of the fastest-appreciating segments of the technology market, with both companies' revenue trajectories so far keeping pace with the rising price tags their investors have assigned.