Loading...
5 tracked companies leading llm inference providers in 2026, by publicly observable engineering signals.
Inference providers are the picks-and-shovels of the AI gold rush — serving open-weight and closed models at scale, often with sub-second time-to-first-token. The category bifurcated in 2025-2026: specialist hardware bets (Groq's LPU, Cerebras) versus GPU-aggregator runtimes.
ai-infra · series c · Python / TypeScript
A quantitative view of Groq's public engineering activity — what we track and why investors watch it.
ai-infra · series b · Python / Rust
A quantitative view of Fireworks AI's public engineering activity — what we track and why investors watch it.
ai-infra · series b · Python
A quantitative view of Together AI's public engineering activity — what we track and why investors watch it.
ai-infra · series b · Python / Go
A quantitative view of Replicate's public engineering activity — what we track and why investors watch it.
ai-infra · series b · Python
A quantitative view of Modal's public engineering activity — what we track and why investors watch it.
Inference is the layer with the clearest infrastructure-engineering signal: GPU-region rollouts, kernel-level optimizations, and continuous model-warming pipelines all show up as commit-velocity bursts and contributor influx tied to new SKUs. Corp Dev at hyperscalers tracks this category for picks-and-shovels acquisitions; emerging-manager funds use it as a sector benchmark.
The signal that matters most: how fast a provider's repo cadence reacts when a new open-weight checkpoint drops from Meta, Mistral, or DeepSeek. Providers that can serve a new SOTA checkpoint within 24-48 hours of release ship a distinctive repo-creation pulse — that pattern is the cleanest leading indicator we publish for this sector.
Compute, orchestration, inference, and the serving layer underneath the model providers. A single page mapping who builds, who funds, and who leads in ai infrastructure.
Frontier labs, model providers, open-weight checkpoints, and the applied-AI layer on top. A single page mapping who builds, who funds, and who leads in ai & machine learning.
Edge platforms, runtimes, networking, observability primitives, and the platform-as-a-service layer. A single page mapping who builds, who funds, and who leads in cloud infrastructure.
From the VC Deal Flow Signal tracked set, the leaders are Groq, Fireworks AI, Together AI, Replicate, Modal. Ranking is by publicly observable engineering acceleration (commit velocity, contributor influx, repo creation pulse, language-bias drift) — not by revenue, valuation, or fundraise size.
Inference is the layer with the clearest infrastructure-engineering signal: GPU-region rollouts, kernel-level optimizations, and continuous model-warming pipelines all show up as commit-velocity bursts and contributor influx tied to new SKUs. Corp Dev at hyperscalers tracks this category for picks-and-shovels acquisitions; emerging-manager funds use it as a sector benchmark.
Companies in the trend are members of the curated /signal/ corpus. The category fit is editorial — companies are included where their public GitHub org clearly ships in this category. Ordering favors the publicly self-described category leader followed by peers ordered by editorial relevance, not by a quantitative score.
The signal that matters most: how fast a provider's repo cadence reacts when a new open-weight checkpoint drops from Meta, Mistral, or DeepSeek. Providers that can serve a new SOTA checkpoint within 24-48 hours of release ship a distinctive repo-creation pulse — that pattern is the cleanest leading indicator we publish for this sector.
Each /signal/[company] page links the underlying GitHub org and the public signal panel. For the full methodology see /methodology and SSRN 6606558. Raw aggregates ship via the public MCP server at /api/v1.
The free Acceleration Watch: five venture-backed teams accelerating on the engineering signal, translated into plain English — 21 to 47 days before the deck circulates. No code-reading, no card.