As of November 16, 2026, Fal is valued at $16 billion after a $1.3 billion Series F led by Bessemer Venture Partners, and its annualized revenue has passed $800 million, per Bloomberg. The San Francisco company, which runs image, video and audio AI models for developers through one API, was worth $1.5 billion just 16 months earlier.
Fal announced the Series F on November 3, 2026 together with a previously undisclosed Series E, saying the two rounds raised $1.65 billion combined. We covered the round itself in our Fal valuation breakdown and the Pulse story on the Series F. This deep dive pulls the whole company into one place: the product, the pricing, the people, the rivals and the case for and against the price.

Fal Company Snapshot
Fal is a generative-media inference company founded in 2021 and based in San Francisco, legally named Features & Labels Inc. It hosts more than 1,000 image, video and audio models behind a single API, had 185 employees as of November 2026, and has raised more than $1.9 billion, with its latest round priced at a $16 billion valuation.
Sources: SiliconANGLE and TechCrunch, November 3, 2026 (citing Bloomberg); Fal blog, March 2026 (model count).
What Does Fal Actually Do?
Fal runs other companies' generative AI models in production so developers don't have to rent, configure and keep GPUs busy themselves. A developer building a video ad tool, a photo editor or a game asset pipeline calls Fal's API, picks a model such as FLUX, Kling, Seedream or Nano Banana, and gets the output back in seconds. Fal's pitch is speed: its engineers tune each model's serving stack so the same weights run faster on Fal than on the model developer's own endpoint.
The niche matters. Most of Fal's well-funded rivals serve text models, where the unit of work is a token. Fal built for media, where the unit is an image or a second of video, and where a single request can tie up a GPU for many seconds. In its May 2026 post announcing a strategic partnership with AWS, Fal said more than 2.5 million developers use the platform and named Amazon MGM Studios, Canva and Adobe as companies running production workloads on it.
Customers, per SiliconANGLE, include entertainment studios, advertising agencies and game developers. Earlier named users include Shopify and Quora, cited in TechCrunch's December 2025 report on the Series D, and Perplexity, named when Fal raised its Series B.
How Does Fal Make Money?
Fal makes money from usage: customers pay for each output a model generates, or for each second of GPU time they reserve. That ties revenue directly to how much media its customers produce, which is why demand spikes translate straight into the top line, and also why compute costs sit directly against every dollar earned.
The two main revenue lines, from Fal's public pricing page as of November 2026:
| Product | Unit | List price |
|---|---|---|
| Model API: FLUX.1 [schnell] image | per megapixel | $0.003 |
| Model API: Seedream V4.5 image | per image | $0.04 |
| Model API: Nano Banana Pro image | per image | $0.15 |
| Model API: Kling Video v3 Pro | per second of video | $0.14 |
| Serverless GPU: H100 (80GB) | per GPU-hour | $4.50 (as low as $2.49) |
| Serverless GPU: H200 (141GB) | per GPU-hour | $6.00 (as low as $2.99) |
| Serverless GPU: B200 (192GB) | per GPU-hour | $7.99 (as low as $5.49) |
How we counted: a sample of 4 Model API prices and all 3 listed Serverless GPU tiers; 7 rows; list prices from fal.ai/pricing as of November 16, 2026. Enterprise and custom deployment pricing is not published.
A third, smaller stream comes from model developers. Some model deals pay Fal licensing fees or a share of revenue, SiliconANGLE reported in November 2026. Fal has not broken out how revenue splits across these lines.
How Much Has Fal Raised? Funding History
Fal has raised more than $1.9 billion across seven rounds since 2021, and four of those rounds came in the 16 months from July 2025 to November 2026. Before its Series B, Fal had raised about $23 million: the February 2025 Series B announcement put total funding at $72 million including that $49 million round.
| Round | Date | Amount | Valuation | Lead(s) |
|---|---|---|---|---|
| Seed + Series A | 2021-2024 | ~$23M combined | Not disclosed | Not confirmed by Fal |
| Series B | Feb 2025 | $49M | Not disclosed | Notable Capital, Andreessen Horowitz |
| Series C | Jul 2025 | $125M | $1.5B | Meritech |
| Series D | Dec 2025 | $140M primary | $4.5B | Sequoia Capital |
| Series E | Apr 2026 (disclosed Nov 2026) | $350M | $8B | Sequoia Capital |
| Series F | Nov 2026 | $1.3B | $16B | Bessemer Venture Partners |
How we counted: every Fal equity round disclosed by the company or reported by Bloomberg, Reuters or TechCrunch; seed and Series A combined because Fal has not published their separate terms; 6 rows; as of November 16, 2026. Sources: Fal blog (Series B, Series C), Reuters, TechCrunch, Tech.eu, SiliconANGLE.
The July 2025 Series C, reported by Reuters, made Fal a unicorn and added Salesforce Ventures, Shopify Ventures and the Google AI Futures Fund. The December 2025 Series D at $4.5 billion also included a secondary sale by existing investors, which is why TechCrunch had first reported a round near $250 million. The April 2026 Series E stayed private until Fal disclosed it alongside the Series F.
Fal Valuation by Round, Jul 2025 - Nov 2026 ($B)
Fal's valuation rose from $1.5 billion to $16 billion in 16 months, roughly doubling at each of its last two rounds
Reuters (Jul 2025); TechCrunch/Bloomberg (Dec 2025); Tech.eu and SiliconANGLE (Nov 2026)
Data
| Item | Value |
|---|---|
| Series C (Jul 2025) | $1.5B |
| Series D (Dec 2025) | $4.5B |
| Series E (Apr 2026) | $8B |
| Series F (Nov 2026) | $16B |
Fal Product Portfolio
Fal's product line has three layers, each sold on the same usage meter.
- Model APIs: a catalog of hosted third-party and open-weight models for text-to-image, image editing, text-to-video, image-to-video, speech and audio, priced per output. Partner launches in 2026 included HeyGen's models in March and Inworld TTS-1.5 Max the same month.
- fal Serverless: launched with the Series C in July 2025, it lets teams deploy their own custom or fine-tuned media models on Fal's GPU fleet, billed by the second.
- Developer tooling: a Playground, the fal Agent, and an MCP server (March 2026) that connects AI assistants to the 1,000-plus model catalog.
Fal's Edge: Post-Training Other Labs' Models to Run Faster
Fal's clearest differentiator in 2026 is that it no longer only hosts models; it post-trains them. In August 2026 it released H3 Max, a post-trained version of MiniMax's open-weights H3 video model, which Fal says generates a 5-second clip in under 3 seconds, about 35 times the throughput of MiniMax's official endpoint. Fal said H3 Max ranked first among 12 video models in its own human-preference tests and was trained and served on Nvidia GB200 NVL72 systems.
The same playbook shows up elsewhere: sub-second serving of Ideogram v4 in July 2026, and an October 2026 technical write-up on EAGLE-4 speculative decoding published with rival Fireworks AI. Tech.eu reported that Fal now develops its own models alongside partners' and plans to fund research into new content-creation methods through 2027.
CEO Burkay Gur told SiliconANGLE that pretraining a Fal-owned foundation model is not on the near-term agenda, though he did not rule it out. One read on this: post-training gives Fal a speed and quality edge on open weights without the billion-dollar cost of a frontier training run, but it also puts Fal one step closer to competing with the model labs it serves.
Fal Revenue and Key Metrics
Fal's annualized revenue passed $800 million in 2026, according to Bloomberg reporting cited by TechCrunch on November 3, 2026; Fal has not published the figure itself. Bloomberg had reported revenue above $200 million as of October 2025, which suggests roughly a fourfold increase in about a year. At its July 2025 Series C, Fal said revenue had grown 60 times over the prior 12 months.
Usage grew alongside revenue. Fal reported more than 1 million developers at its February 2025 Series B and 50-plus enterprise customers; by May 2026 it cited more than 2.5 million developers. Demand outran capacity during summer 2026: "We hit a point in summer where we were not able to serve our customers and had to rate-limit them," Gur told SiliconANGLE.
Gross margin is not publicly disclosed. TechCrunch's sources said margins across inference companies remain thin because compute is expensive to buy or lease, and Fal says the Series F will go toward computing power and hiring. At $16 billion on $800 million-plus of annualized revenue, Fal trades at roughly 20 times run-rate revenue (Value Add VC's arithmetic).
Who Are Fal's Competitors?
Fal's closest comparable is Fireworks AI, which announced a $1.505 billion Series D at a $17.5 billion valuation in July 2026 and said annualized revenue had passed $1 billion. Fireworks mostly serves text and code models; Fal serves media. Other rivals for the same inference budgets include Baseten, Together AI and Modal, profiled in our Baseten deep dive, Together AI deep dive and Modal valuation post.
Fal vs Fireworks AI: Valuation and Annualized Revenue, 2026
SiliconANGLE and TechCrunch (Fal, Nov 2026); Fireworks AI blog (Jul 2026)
Fal is priced at about 20x run-rate revenue versus about 17.5x for Fireworks AI (Value Add VC's arithmetic on the disclosed figures).
Model developers are a second, quieter threat. Labs whose models run on Fal, such as MiniMax, also operate their own official endpoints, so a customer can skip the middleman. Fal's defense is speed and breadth: one integration reaches over 1,000 models, and its tuned versions run faster than the originals.
Fal Leadership Team
Fal is run by its two co-founders, whom Tech.eu describes as Turkish entrepreneurs. CEO Burkay Gur previously led machine-learning work at Coinbase and held an engineering role at Oracle; CTO Gorkem Yurtseven was a developer at Amazon. They started the company in 2021 as Features & Labels, a name that still sits on its legal filings.
The board reflects each growth stage. Jennifer Li of Andreessen Horowitz and Glenn Solomon of Notable Capital joined with the Series B, Arsham Memarzadeh of Meritech with the Series C, and Talia Goldberg of Bessemer Venture Partners with the Series F. Sequoia Capital led two rounds in a row, and Nvidia invested in both the Series D and Series F.
Fal Bull Case / Bear Case
Bull case
- Video is the heaviest generative workload per request, and Fal is the specialist built for it.
- Reported revenue went from above $200 million in October 2025 to an $800 million-plus run rate in 2026, and summer 2026 demand exceeded capacity.
- Post-trained models like H3 Max give Fal products rivals cannot simply resell.
- Investors include Nvidia, Sequoia and Bessemer, and the 2026 rounds fund compute well into 2027.
Bear case
- Gross margin is undisclosed, and sector sources describe inference margins as thin.
- About 20x run-rate revenue is a richer multiple than Fireworks AI's at its last round.
- Model labs and hyperscalers can serve the same models directly and compete on price.
- Usage pricing means revenue falls quickly if a large customer moves workloads in-house.
What the headline misses
The $16 billion and $1.3 billion figures come from Bloomberg's reporting; the number Fal itself confirmed is the $1.65 billion combined for the Series E and F. The $800 million revenue figure is also reported, not disclosed. Neither is in doubt from the sources we found, but every multiple in this post rests on reported rather than audited numbers.
Will Fal IPO? IPO Outlook
Fal has announced no IPO as of November 16, 2026: we found no filing, no reported bank mandate and no executive comment on a listing timeline. None of the November 2026 coverage of the Series F, in TechCrunch, SiliconANGLE or Tech.eu, mentions public-market plans.
The signals point the other way for now. A company that can raise $1.65 billion privately in one year, and says it will spend that money on GPUs, hiring and research through 2027, has little need to list. This likely means an IPO is not a near-term question for Fal, and the more useful near-term marker is whether Fal discloses a gross margin at its next raise.
The Bottom Line
Fal turned a narrow bet, that image and video models would need their own inference layer, into a $16 billion company with 185 employees and a revenue run rate past $800 million. The growth is real and the investor list is strong. What is missing is proof that the business keeps enough of each dollar after paying for GPUs, and until Fal shows its margin, the valuation is a bet on volume continuing to outrun compute costs.
All figures are from company announcements and credible reporting as of November 16, 2026; the Series F size, valuation and revenue are as reported by Bloomberg. Track AI infrastructure rounds with Value Add VC's funding tracker and our free Startup Funding Dataset.
Latest from the Pulse
Get VC data most people never see
— free to subscribe
Trace's notes on venture, AI and startups, a few times a week. Join 5,000+ subscribers. No spam.
No spam, unsubscribe anytime. Sent via Substack · Privacy · or subscribe on Substack ↗