Verda raises $189M: another place to run inference
Helsinki's Verda hit unicorn status with a $189M Series B and a $165M revenue run rate. The neocloud tier is now real enough to quote against your hyperscaler bill.
Verda raised $189 million and crossed a $1 billion valuation. The Helsinki company announced an oversubscribed Series B led by Emergence Capital on September 22, Tech.eu reported, bringing total funding past $450 million. The interesting number is not the raise. It is the $165 million annualized revenue run rate Verda hit in July — the challenger AI cloud tier now has customers, not just capital.
What actually happened
Verda was founded in Helsinki in 2020 and runs a full-stack AI cloud: physical data centers and hardware at the bottom, a cloud platform above it, AI research on top. It sells on-demand compute for AI workloads across Europe, the US and Asia, and has opened offices in London and San Francisco.
The round brought in MUFG Innovation Partners, Supermicro, Varma Mutual Pension Insurance Company, Lifeline Ventures, 6 Degrees Capital, byFounders and Tesi, plus angels including Meta's Ola Tørudbakken and Mark Saroufim.
Verda says the money goes to product development across every layer of the stack, with a stated focus on inference, plans to multiply compute capacity over the next year, and continued international expansion.
Why it matters for your business
Most small and mid-size teams never price their inference layer. You picked Bedrock or Vertex or a first-party API because it was there, you wired it in, and the bill became a fixed cost nobody revisits. That was defensible in 2024, when there was no second quote to get.
It is not defensible now. Verda is one of several neoclouds with enough revenue to be a real vendor, and "focus on inference" is a direct shot at the margin hyperscalers earn on token serving. When a challenger tier funds itself specifically to undercut your unit economics, the correct response is to find out what you are actually paying.
The move is not to migrate. It is to stop being unable to migrate:
Know your cost per unit of work. Not monthly spend — cost per support ticket resolved, per product description generated, per document parsed. If you cannot produce that number in a minute, you cannot evaluate any alternative, and every vendor conversation ends with you accepting the renewal.
Keep the model call behind your own interface. One function in your codebase that takes a prompt and returns a completion, with the provider as configuration. We build every client system this way. It costs an afternoon up front and it is the difference between a pricing conversation and a rebuild.
Separate batch work from live inference. Batch jobs are the easiest thing to move to a cheaper provider and the lowest-risk place to test one. Run a week of your overnight enrichment on a second vendor before you consider touching anything a customer waits on.
Capital flowing into challenger compute is good news for buyers — but only for buyers whose architecture lets them act on it. Portability is not a philosophical position. It is leverage with a dollar value.
Key takeaways
- Verda raised a $189M Series B led by Emergence Capital, announced September 22, 2026, at a $1B+ valuation
- Total funding now exceeds $450M; the company hit a $165M annualized revenue run rate in July
- Founded Helsinki 2020; full-stack AI cloud from data centers to platform, serving Europe, the US and Asia
- Stated use of funds includes an explicit focus on inference and multiplying compute capacity within a year
- Track cost per unit of work, not monthly spend — it is the only number that makes a second quote meaningful
- Put every model call behind one internal interface so the provider is configuration, not architecture
A vendor you cannot leave is a vendor that sets your price. We build AI systems with the model layer behind an interface you own, so switching providers is a config change and a cost test — not a rewrite. See how we build vendor-agnostic systems, or put real numbers on your current inference spend.
- #ai-infrastructure
- #inference
- #vendor-pricing
- #gpu-cloud
- #verda
Tommy Rush — Founder, Rush Commerce
Operator turned builder. 15+ years running operations — now shipping the systems businesses run on. More
Get The Rush Report weekly — one email, zero fluff.
Keep reading
AI CEOs brief the UN Security Council: what changes for you
France convened the Security Council's first session on AI safety risks on Sept 23, with Altman, Amodei, Bengio and Delangue briefing. Here's the operator read.
Read itSnorkel AI raises $350M at $3.5B: data work is the moat
Snorkel's revenue grew 18x to a $375M run rate by selling finished datasets, not labeling software. The lesson for small studios is about what you sell.
Read it