AI niche map

AI infrastructure companies: 36 funded since September 27, 2025, updated October 14, 2026

Started September 27, 2026 · Updated October 14, 2026 · Presspool

Presspool has verified 36 AI infrastructure companies that announced a priced round since September 27, 2025. In its latest weekly sample of AI answers, through October 14, 2026, the ones AI apps named or cited most were Together AI (16 of 50 answers), Lambda (15 of 50 answers) and GMI Cloud (12 of 50 answers). Their latest announced rounds total $17.59 billion.

Companies selling the layers other AI products run on: GPU clouds and neoclouds, inference and model-serving platforms, training data and web data for AI, evaluation, observability and testing for AI, vector and AI-native databases, and infrastructure for AI agents. Each entry is taken from the company's own website and confirmed by an independent source, both linked on the entry. No company pays to be listed.

CompanyWhat it doesLatest announced financingCited in the Machine Relations Index
CrusoeCrusoe is a vertically integrated AI infrastructure provider that builds its own AI factories and runs Crusoe Cloud on top of them. (site, source)$3.9 billion, Series F (initial closing) (September 17, 2026)Cited in 2 of 25,323 Index answers, May 10, 2026 to October 16, 2026
NscaleNscale is a full-stack AI cloud platform spanning software, compute and power, from AI training and inference to data centers. (site, source)$3.36 billion, Pre-IPO convertible notes (September 25, 2026)Cited in 4 of 25,323 Index answers, May 10, 2026 to October 16, 2026
Fireworks AIFireworks is a platform where companies train, customize and serve open models, with an inference engine tuned for throughput and latency. (site, source)$1.505 billion, Series D (July 16, 2026)Cited in 9 of 25,323 Index answers, May 10, 2026 to October 16, 2026
BasetenBaseten is an AI inference company that post-trains and serves customers' models at production scale, with a focus on latency, reliability and cost. (site, source)$1.5 billion, Series F (June 22, 2026)Cited in 1 of 25,323 Index answers, May 10, 2026 to October 16, 2026
LambdaLambda is an AI cloud, which it calls the Superintelligence Cloud, that provides GPU supercomputers for training and inference. (site, source)over $1.5B, Series E (November 18, 2025)Cited in 10 of 25,323 Index answers, May 10, 2026 to October 16, 2026
FluidstackFluidstack builds and deploys large-scale compute infrastructure for leading AI labs. (site, source)$830M, Series A (July 20, 2026)Cited in 0 of 25,323 Index answers, May 10, 2026 to October 16, 2026
Together AITogether AI runs an AI-native cloud: a full-stack platform for inference, fine-tuning and GPU clusters built around open models. (site, source)$800 million, Series C (July 1, 2026)Cited in 8 of 25,323 Index answers, May 10, 2026 to October 16, 2026
ModalModal is a cloud built for AI, giving developers primitives for workloads such as low-latency elastic inference. (site, source)$355 million, Series C (May 21, 2026)Cited in 19 of 25,323 Index answers, May 10, 2026 to October 16, 2026
GroqGroq runs an AI inference cloud, a neocloud serving open models on its own data-center capacity, after licensing its chip technology to Nvidia. (site, source)$350 million, Series A (August 17, 2026)Cited in 1 of 22,910 Index answers, May 10, 2026 to October 16, 2026
MercorMercor organizes human experts to power the AI economy, offering expert work for AI labs, custom evals and the APEX benchmarks of frontier-model performance. (site, source)$350M, Series C (October 27, 2025)Cited in 0 of 25,323 Index answers, May 10, 2026 to October 16, 2026
Snorkel AISnorkel AI builds expert-curated datasets and environments that AI labs, enterprises and government agencies use to train and evaluate models. (site, source)$350 million, Series E (September 22, 2026)Cited in 14 of 25,323 Index answers, May 10, 2026 to October 16, 2026
TensorWaveTensorWave is an all-AMD AI cloud for high-performance, memory-intensive training and inference workloads. (site, source)$350 million, Series B (June 10, 2026)Cited in 0 of 25,323 Index answers, May 10, 2026 to October 16, 2026
SiliconFlowSiliconFlow runs an AI inference cloud that serves and fine-tunes more than 200 language and multimodal models through APIs. (site, source)more than 2 billion yuan ($294 million), Series B (June 17, 2026)Cited in 26 of 25,826 Index answers, May 10, 2026 to October 16, 2026
ExaExa offers search over a large index of public web and private information, delivered through APIs built for AI agents. (site, source)$250M, Series C (May 20, 2026)Cited in 1 of 25,323 Index answers, May 10, 2026 to October 16, 2026
GMI CloudGMI Cloud is a neocloud selling NVIDIA GPU clusters, bare metal, and serverless and dedicated inference. (site, source)$223 million, Series B (September 30, 2026)Cited in 116 of 25,826 Index answers, May 10, 2026 to October 16, 2026
VerdaVerda, formerly DataCrunch, is a full-stack AI cloud company that runs its own data centers, hardware and platform to provide on-demand compute for AI. (site, source)$189 million (€163 million), Series B plus additional investment (September 22, 2026)Cited in 6 of 25,323 Index answers, May 10, 2026 to October 16, 2026
InferactInferact was founded by the creators and core maintainers of vLLM to grow it as an AI inference engine and make inference cheaper and faster. (site, source)$150 million, Seed (January 22, 2026)Cited in 0 of 25,323 Index answers, May 10, 2026 to October 16, 2026
LMArenaArena runs a public AI ranking and LLM leaderboard and positions itself as an evaluation platform for frontier AI. (site, source)$150M, Series A (January 6, 2026)Cited in 0 of 25,323 Index answers, May 10, 2026 to October 16, 2026
falfal is a generative media platform that runs 1,000+ image, video, audio and 3D models through APIs, serverless deployment and GPU compute. (site, source)$140 million, Series D (December 9, 2025)Cited in 1 of 25,323 Index answers, May 10, 2026 to October 16, 2026
LangChainLangChain builds a platform for agent engineering, spanning its open-source agent frameworks and the LangSmith platform for testing and improving agents with production data. (site, source)$125M, Series B (October 20, 2025)Cited in 32 of 25,323 Index answers, May 10, 2026 to October 16, 2026
DeepInfraDeepInfra runs an inference cloud that serves 150+ open-source models through OpenAI-compatible APIs on GPU infrastructure it owns and operates. (site, source)$107 million, Series B (May 4, 2026)Cited in 4 of 25,323 Index answers, May 10, 2026 to October 16, 2026
Parallel Web SystemsParallel builds web infrastructure that lets AI systems search, extract, monitor and reason over information on the web. (site, source)$100M, Series B (April 29, 2026)Cited in 0 of 25,323 Index answers, May 10, 2026 to October 16, 2026
RadixArkRadixArk, founded by the creators of the SGLang inference engine, builds open AI infrastructure for inference and reinforcement learning, including managed infrastructure for teams. (site, source)$100 million, Seed (May 5, 2026)Cited in 0 of 25,323 Index answers, May 10, 2026 to October 16, 2026
RunpodRunpod is an AI developer cloud where developers build, train, fine-tune, deploy and scale AI workloads. (site, source)$100 million, Series A (June 24, 2026)Cited in 99 of 25,323 Index answers, May 10, 2026 to October 16, 2026
BraintrustBraintrust is an observability and evals platform for AI agents that traces production behavior and lets teams run experiments against datasets before shipping. (site, source)$80M, Series B (February 17, 2026)Cited in 35 of 25,323 Index answers, May 10, 2026 to October 16, 2026
FirecrawlFirecrawl is a web data API that searches, scrapes and interacts with websites, turning them into clean Markdown or structured data for AI agents. (site, source)$75 million, Series B (September 22, 2026)Cited in 62 of 25,323 Index answers, May 10, 2026 to October 16, 2026
OllamaOllama is an API for open models that runs inference locally or in the cloud. (site, source)$65 million, Series B (July 9, 2026)Cited in 0 of 25,323 Index answers, May 10, 2026 to October 16, 2026
ArcadeArcade is an actions runtime that sits between enterprise AI agents and business systems to authorize, execute and govern every agent action. (site, source)$60M, Series A (June 15, 2026)Cited in 0 of 25,323 Index answers, May 10, 2026 to October 16, 2026
Patronus AIPatronus AI trains Digital World Models that simulate agent actions in digital workflows to generate simulations frontier models can train and be evaluated on. (site, source)$50M, Series B (June 25, 2026)Cited in 6 of 25,323 Index answers, May 10, 2026 to October 16, 2026
QdrantQdrant is a vector search engine for building high-performance AI retrieval at any scale and under any deployment model. (site, source)$50M, Series B (March 12, 2026)Cited in 20 of 25,323 Index answers, May 10, 2026 to October 16, 2026
ParasailParasail is an inference cloud that serves 40+ open and frontier models through one OpenAI-compatible API and tunes deployments for speed, quality and cost. (site, source)$32 million, Series A (April 15, 2026)Cited in 15 of 25,323 Index answers, May 10, 2026 to October 16, 2026
Fiddler AIFiddler is an enterprise AI observability and security platform positioned as a control plane for monitoring, evaluating and governing AI agents. (site, source)$30M, Series C (January 27, 2026)Cited in 11 of 25,323 Index answers, May 10, 2026 to October 16, 2026
DaytonaDaytona provides secure, elastic sandbox infrastructure for running AI-generated code, with fast environment creation and stateful operation. (site, source)$24M, Series A (February 5, 2026)Cited in 0 of 25,323 Index answers, May 10, 2026 to October 16, 2026
KernelKernel gives web agents cloud browser infrastructure, with isolated Chromium sessions controlled via API or SDK plus live view, replays and telemetry. (site, source)$22M, Seed and Series A (combined) (October 9, 2025)Cited in 0 of 25,323 Index answers, May 10, 2026 to October 16, 2026
HUDHUD is a platform for building reinforcement-learning environments that encode domain expertise to train and evaluate models and produce post-training data. (site, source)$16M, Series A (June 30, 2026)Cited in 0 of 25,323 Index answers, May 10, 2026 to October 16, 2026
Thunder ComputeThunder Compute sells virtualized cloud GPU instances and GPU sandboxes for AI development and agents. (site, source)$13 million, Series A (August 19, 2026)Cited in 24 of 25,826 Index answers, May 10, 2026 to October 16, 2026

Which of these companies AI names

On October 14, 2026, Presspool asked the 8 questions below on ChatGPT, Claude, Perplexity, Gemini, Copilot, Google AI Mode and Google AI Overviews and recorded 50 answers. Not sampled: 1 Google AI Mode answer, 5 Google AI Overviews answers. A company counts when an answer names it or cites its website. Answers vary from run to run, so each count is a dated sample, not a verdict.

CompanyNamed or cited inChatGPTClaudePerplexityGeminiCopilotGoogle AI ModeGoogle AI Overviews
Crusoe10 of 50 answers3 of 80 of 83 of 82 of 81 of 80 of 71 of 3
Nscale5 of 50 answers2 of 80 of 82 of 80 of 81 of 80 of 70 of 3
Fireworks AI10 of 50 answers2 of 81 of 82 of 81 of 81 of 81 of 72 of 3
Baseten5 of 50 answers2 of 80 of 81 of 81 of 80 of 80 of 71 of 3
Lambda15 of 50 answers3 of 81 of 83 of 83 of 82 of 82 of 71 of 3
Fluidstack5 of 50 answers1 of 81 of 83 of 80 of 80 of 80 of 70 of 3
Together AI16 of 50 answers2 of 82 of 82 of 82 of 83 of 83 of 72 of 3
Modal11 of 50 answers2 of 81 of 83 of 82 of 81 of 82 of 70 of 3
Groq8 of 50 answers1 of 80 of 82 of 81 of 82 of 81 of 71 of 3
Mercor0 of 50 answers0 of 80 of 80 of 80 of 80 of 80 of 70 of 3
Snorkel AI0 of 50 answers0 of 80 of 80 of 80 of 80 of 80 of 70 of 3
TensorWave0 of 50 answers0 of 80 of 80 of 80 of 80 of 80 of 70 of 3
SiliconFlow3 of 50 answers0 of 80 of 80 of 81 of 80 of 81 of 71 of 3
Exa3 of 50 answers1 of 80 of 80 of 81 of 80 of 81 of 70 of 3
GMI Cloud12 of 50 answers2 of 82 of 83 of 83 of 80 of 81 of 71 of 3
Verda1 of 50 answers1 of 80 of 80 of 80 of 80 of 80 of 70 of 3
Inferact1 of 50 answers1 of 80 of 80 of 80 of 80 of 80 of 70 of 3
LMArena3 of 50 answers0 of 80 of 80 of 82 of 81 of 80 of 70 of 3
fal2 of 50 answers1 of 80 of 80 of 80 of 80 of 81 of 70 of 3
LangChain7 of 50 answers1 of 81 of 81 of 81 of 81 of 82 of 70 of 3
DeepInfra2 of 50 answers1 of 80 of 80 of 80 of 80 of 81 of 70 of 3
Parallel Web Systems2 of 50 answers1 of 80 of 80 of 81 of 80 of 80 of 70 of 3
RadixArk0 of 50 answers0 of 80 of 80 of 80 of 80 of 80 of 70 of 3
Runpod4 of 50 answers3 of 80 of 81 of 80 of 80 of 80 of 70 of 3
Braintrust6 of 50 answers1 of 81 of 81 of 81 of 81 of 81 of 70 of 3
Firecrawl10 of 50 answers1 of 83 of 81 of 81 of 80 of 84 of 70 of 3
Ollama4 of 50 answers1 of 80 of 80 of 81 of 81 of 81 of 70 of 3
Arcade0 of 50 answers0 of 80 of 80 of 80 of 80 of 80 of 70 of 3
Patronus AI0 of 50 answers0 of 80 of 80 of 80 of 80 of 80 of 70 of 3
Qdrant1 of 50 answers0 of 80 of 80 of 81 of 80 of 80 of 70 of 3
Parasail0 of 50 answers0 of 80 of 80 of 80 of 80 of 80 of 70 of 3
Fiddler AI0 of 50 answers0 of 80 of 80 of 80 of 80 of 80 of 70 of 3
Daytona5 of 50 answers1 of 80 of 80 of 81 of 81 of 82 of 70 of 3
Kernel1 of 50 answers0 of 80 of 80 of 80 of 80 of 81 of 70 of 3
HUD0 of 50 answers0 of 80 of 80 of 80 of 80 of 80 of 70 of 3
Thunder Compute1 of 50 answers0 of 80 of 80 of 80 of 80 of 81 of 70 of 3

What changed since the week of October 5, 2026

Sources the AI apps cited most for these questions

  1. openai.com: 8 of 50 answers
  2. gmicloud.ai: 7 of 50 answers
  3. builtin.com: 6 of 50 answers
  4. digitalocean.com: 6 of 50 answers
  5. firecrawl.dev: 6 of 50 answers
  6. github.com: 6 of 50 answers
  7. landbase.com: 6 of 50 answers
  8. crunchbase.com: 5 of 50 answers
  9. mastra.ai: 5 of 50 answers
  10. modal.com: 5 of 50 answers
  11. newmarketpitch.com: 5 of 50 answers
  12. youtube.com: 5 of 50 answers
  13. fireworks.ai: 4 of 50 answers
  14. northflank.com: 4 of 50 answers
  15. reddit.com: 4 of 50 answers

What counts as AI infrastructure

A company joins this map when its own site describes AI infrastructure as its main product and an independent source confirms it, and it publicly announced a priced funding round in the last 12 months. The table shows each company's most recent announced financing, which can be a later instrument such as convertible notes; the qualifying round is in the data file. Out of scope: companies whose product is an end-user AI application, and companies whose most recent announced round is more than 12 months old. The Machine Relations Index column counts the answers in the Machine Relations Index that cited the company's website; the Index asks a fixed panel of buyer questions about software and services, so it measures how often AI engines already treat the company as a source. Definitions are versioned; this is version 1. Corrections need a public source: see Contact.

Questions this map tracks

Changes