Fastest inference platform for generative AI models.
Public company, workplace, funding, and market signals
Updated Aug 2, 2026
Fireworks AI is a privately held AI-infrastructure company founded by former Meta PyTorch and Google Vertex AI engineers. It helps companies train, specialize, deploy, and serve open-weight and proprietary models with high throughput, low latency, and lower cost. Its current positioning is enabling companies to own “specialized intelligence” built on their own data and workflows.
Primary product
A production AI platform combining Fireworks Training and Fireworks Inference: guided/configurable/custom fine-tuning, model hosting, serverless inference, dedicated/on-demand GPU deployments, reserved capacity, and OpenAI-/Anthropic-compatible APIs.
Founded
2022
Headquarters
San Mateo, California, United States
Team size
Approximately 200 as of July 2026; the company has publicly discussed growing to approximately 600 by year-end 2026.
Work style
Flexible
Industry
0 jobs at Fireworks AI
Check back later for new openings
Software Development / Artificial Intelligence Infrastructure
Sub-industry
Generative-AI model training, inference, and cloud GPU infrastructure
Offices
Business model
Stage
Series D / late-stage private company
Total raised
$1.8B
Latest round
Series D · Jul 2026
Latest amount
$1.5B
Jul 2026 · Atreides Management, Index Ventures, and TCV
Oct 2025 · Lightspeed Venture Partners, Index Ventures, and Evantic
Jul 2024 · Sequoia Capital
Mar 2024 · Benchmark
Jan 2022 · Benchmark
Leadership
Co-Founder and CEO
Co-Founder
Co-Founder
Co-Founder
Co-Founder
Co-Founder
Co-Founder
President
Founders
Co-Founder
Investors
Public evidence points to an engineering-led, technically intense, fast-moving environment built around high autonomy, rapid shipping, infrastructure performance, and cross-functional ownership. The founding team has deep PyTorch/Meta and Google Vertex AI backgrounds. Employee-review evidence is limited and mixed; it describes strong technical peers and a demanding pace. The company does not publish a complete benefits or values package on the researched public pages.
Work style
Flexible
Visa sponsorship
Limited
Compensation
Public job and H-1B/LCA data provide signals rather than an official compensation framework: reported sponsored-role salaries have a median around $220,000, with technical roles commonly showing approximately $145,000-$350,000 base ranges depending on level and location. Some public sales listings show approximately $200,000-$240,000 OTE plus equity. Equity and exact benefits are not publicly standardized.
Pricing
Pay per input/output token for serverless inference; per-token or GPU-hour pricing for training; per-GPU-second/hour pricing for on-demand deployments; reserved-capacity and enterprise arrangements for guaranteed capacity and quotas.
Differentiators
Technology
Customers
Competitors
Estimated revenue
Approximately $1.0B-$1.5B annualized revenue run rate as of July 2026. Fireworks publicly reported over $1B ARR; the upper bound is a research range, not a reported figure.
Estimated monthly visits
542K
Fireworks acquired Hathora to add global compute orchestration and low-latency routing expertise. Hathora wound down or migrated its gaming-infrastructure customer business, making this both a capability acquisition and a product-scope transition.
SourceCNBC reported that roughly half of Fireworks revenue had come from Cursor in the prior year, although CEO Lin Qiao said the company was more diversified by July 2026. Customer concentration is therefore a material historical risk signal.
SourceThe company is scaling rapidly after its Series D, with approximately 200 employees and a stated goal of approximately 600 by year-end 2026; this creates execution, hiring, infrastructure, and integration risk, though no layoffs or shutdown were found.
SourceThe Next Web · Jul 2026
The article frames Fireworks' strategy as helping enterprises build specialized intelligence with open models at materially lower cost than frontier APIs, while identifying Together AI and Baseten as inference competitors.
CNBC · Jul 2026
CNBC reported the funding, approximately 200 employees, expected hiring growth, George Hu joining as president, customer concentration history, and competition with hyperscalers and inference startups.
Fireworks AI · Jul 2026
Fireworks announced a $1.505B Series D at a $17.5B valuation, reported over $1B ARR and more than 40T tokens processed daily, and disclosed the Hathora acquisition.
George Hu / LinkedIn · Apr 2026
George Hu announced that he joined Fireworks as President; Lin Qiao described the appointment as a step toward scaling the company commercially.
SiliconANGLE · Mar 2026
Follow-up coverage explained that Hathora was acquired primarily for talent and global compute-orchestration infrastructure supporting low-latency inference and training.