Silurian

What do they actually do

Silurian provides a live weather API (the Earth API) that serves hourly, location‑specific forecasts and renewable‑energy variables. Developers can try it in a web playground and integrate it via Python/TypeScript SDKs and REST endpoints Earth API announcement API docs.

Under the hood, they train large pretrained weather models (the GFT family) and then post‑train them on customer sensor/asset data to produce higher‑resolution, asset‑aware forecasts and risk signals (e.g., hub‑height winds, icing risk) YC listing Hydro‑Québec study. They’ve shown public demos and a U.S. regional model (GFT‑US) that is live and “free to try” for early users Earth API announcement GFT‑US launch.

A named partner is Hydro‑Québec, where Silurian post‑trained its model on utility sensor streams to generate asset‑level forecasts and operational signals for grid decisions Hydro‑Québec study. The company is still early commercially and focused on pilots, expanding the API surface, and productizing the post‑training workflow rather than broad enterprise rollout today Earth API announcement YC listing.

Who are their target customer(s)

• Electric utilities and transmission operators: Need reliable, site‑specific forecasts to reduce outages and plan day‑ahead dispatch, especially where icing and hub‑height winds damage assets; current tools often miss asset‑level effects Hydro‑Québec study.
• Wind and solar plant operators: Struggle to predict short‑term production and turbine issues (e.g., icing) that drive curtailments and maintenance; asset‑tuned hourly forecasts can improve scheduling and bids Earth API.
• Energy traders and schedulers: Lose money when forecasts miss hour‑to‑hour variability at specific sites and heights; need more precise, probabilistic forecasts aligned to trading and dispatch horizons Grid blog.
• Insurers and reinsurers: Require localized hazard and loss‑probability estimates to price risk and settle claims; coarse or generic products leave large exposure uncertainty YC listing.
• Agriculture producers and logistics/fleet operators: Weather variability drives crop loss, spoilage, and route disruptions; want hourly field‑ or asset‑level forecasts to automate irrigation, harvest timing, or re‑routing Earth API.

How would they acquire their first 10, 50, and 100 customers

• First 10: Run targeted pilots with large asset owners: use the public playground and “free to try” regional model to validate baseline accuracy, then post‑train on the customer’s sensor feed to deliver one or two asset‑level signals and co‑publish a brief (as with Hydro‑Québec) Earth API announcement Hydro‑Québec study.
• First 50: Package a fixed‑scope onboarding with template data connectors and a playbook that delivers a post‑trained model plus validated operational thresholds; reuse early pilots as case studies and referrals to shorten sales cycles GFT roadmap Hydro‑Québec case study.
• First 100: Scale via channel partners (SCADA/asset‑management vendors, energy software platforms, SIs) while offering a clear self‑serve API/SDK for smaller operators; drive inbound with technical content, demos, and Aurora/Nature credibility plus published case studies API docs Aurora paper.

What is the rough total addressable market

Top-down context:

Covers buyers of weather APIs and operational decision support across energy, utilities, insurance, agriculture, and logistics—where weather materially affects revenues, costs, and risk. Silurian’s near‑term focus is the energy and utilities beachhead, expanding to adjacent sectors as the post‑training workflow is productized.

Bottom-up calculation:

Illustrative beachhead TAM: utilities (100 large T&D operators × $200k/year) ≈ $20M; utility‑scale wind/solar sites (2,000 sites × $30k/year) ≈ $60M; energy trading firms (200 × $50k/year) ≈ $10M; insurers/reinsurers (50 × $150k/year) ≈ $7.5M; ag/logistics enterprises (1,000 × $20k/year) ≈ $20M. Combined initial opportunity ≈ $115M–$130M, with upside from multi‑site rollouts and additional decision outputs.

Assumptions:

• ACVs reflect a post‑trained model plus 1–2 operational signals; API‑only usage is lower and not included.
• Counts are conservative, focusing on large operators and utility‑scale sites; excludes long tail and other regions.
• Does not include expansion revenue from additional assets, geographies, or non‑weather Earth‑system outputs.

Who are some of their notable competitors

• Tomorrow.io: Developer‑friendly weather API and enterprise “weather intelligence” platform sold into energy, utilities, and logistics; overlaps on customers and API delivery but emphasizes an operational platform with proprietary sensing rather than customer post‑trained foundation models.
• The Weather Company / IBM: Longstanding provider of weather data, APIs, and enterprise solutions (including outage prediction) with deep SLAs and integrations; differs from Silurian’s model‑centric, customer post‑training approach by being a legacy full‑stack provider within IBM’s software ecosystem.
• Meteomatics: High‑resolution weather models and API focused on renewable‑energy forecasting and downscaling; overlaps on energy use cases but emphasizes tailored high‑res modeling and services rather than a foundation‑model fine‑tuning workflow.
• Vaisala: Hardware + software for wind/solar operators (lidar, sensors, site‑specific forecasting/monitoring); more sensor‑ and instrumentation‑first, often complementary but a different go‑to‑market than Silurian’s model‑centric approach.
• DTN: Sector‑focused weather platform with decisioning products for agriculture, utilities, logistics, and energy, supported by meteorologists; overlaps on operational decision support but is a human+platform service rather than customer‑fine‑tuned foundation models.