Venice going vertical: B300 clusters

What happened

Lead with the official raise. On July 1, 2026, Venice’s blog announced a $65M Series A led by Dragonfly at a $1B valuation (its first outside capital) with Coinbase Ventures, F-Prime, North Island Ventures among backers. The same disclosure put usage near ~3.5M users and ~1.3T tokens processed per month. That post is the announcement; later tweets are execution updates.

On July 24, CTO Jesse Proudman posted a progress checklist (“Series A: secured. GPUs: on the way. Data centers: deploying…”), not a second fundraise headline. Network infrastructure hiring ran alongside it.

On July 29, Jesse posted that Venice is going vertical: standing up B300 GPU clusters in Venice’s own data centers, and recruiting people who live in CUDA kernels, quantization math, and production inference at scale, including hard experience with vLLM, SGLang, Nsight, and tensor-parallel multi-node configs, for privacy-first infrastructure for millions of users. Port City of AI as metal and software, not only a frontend. The posting itself confirms the plan: “Help stand up and optimize Venice's owned GPU infrastructure, including B300 nodes in our data centers.”

Jesse Proudman on X: Series A secured, GPUs on the way, data centers deploying — Jesse Proudman (Jul 24, 2026): progress checklist after the Series A announcement + network engineer hiring.

Jesse Proudman on X: Venice is going vertical, B300 clusters and inference hiring — Jesse Proudman (Jul 29, 2026): B300 clusters in own data centers + production-inference hiring.

Timeline: Jul 1 → 7 → 24 → 27 → 29

July 1, 2026

Series A announced: $65M led by Dragonfly at $1B valuation; first outside capital (Coinbase Ventures, F-Prime, North Island Ventures among backers). Discloses ~3.5M users and ~1.3T tokens processed/month. Official source: Venice blog.

July 7, 2026

Venice registers ARIN org record VENIC-4. Network infrastructure prep begins six days after the raise. Speed of execution; ARIN prep did not predate the funding announcement.

July 24, 2026

Jesse Proudman: “Series A: secured. GPUs: on the way. Data centers: deploying…” + Senior Network Infrastructure Engineer posting (~44K views). Checklist, not the Series A announcement.

July 27, 2026

AS402818 (VENICE-NET-1) registered with ARIN; Jesse posts “It’s happening…” with whois (~21K views). Read the ASN note →

July 29, 2026

“Venice is going vertical”: B300 GPU clusters in own data centers + CUDA / vLLM / SGLang hiring (~4.6K views).

What it might mean

Own metal: B300-class clusters in Venice data centers = capital and ops commitment to first-party capacity, not only marketplace GPUs.
Scale justification (margin, not only ideology): at roughly 1.3T tokens/month and peaks around ~300K inference requests/hour, the choice between owning B300 capacity and forever-renting is a unit-economics and reliability decision at production load. Privacy still matters; the volume is why vertical integration shows up in the roadmap after a real raise.
Inference quality and cost: teams that push vLLM/SGLang and multi-node parallelism optimize latency, throughput, and $/token, the guts of API and chat capacity. The inference posting names the goal directly: drive down latency and improve cost per token at scale.
Privacy stack: more of the path on infrastructure Venice controls fits fewer intermediaries and stronger privacy guarantees.
Same arc as the ASN: network (Jul 27) + compute (Jul 29) follows Jul 1 capital and Jul 7 ARIN org prep. One vertical sequence.

What it does NOT mean

Not an automatic DIEM mint-rate drop.
Not guaranteed APR, emissions, or token price support.
Not proof that every query already runs only on Venice-owned metal end-to-end.
Not a reason to skip mint vs buy math — infra news ≠ free mint premium.
Not the same thing as the Jul 17 tokenomics update — programmatic VVV burns and the DIEM supply-target path are protocol mechanics; they exist independently of the infrastructure build.

For holders and users

Bullish infra is not a substitute for acquisition math. Mint rate, DIEM market price, locked-VVV yield (~6.77% APR in this guide’s current assumptions), and unlock risk still decide whether minting beats buying.

Use this note as context for who Venice is becoming operationally. Use the calculators for what you should do with capital today.

If owned infrastructure lowers inference costs and grows API usage, that flows into Venice's existing programmatic VVV burn and staking emissions.

Both job postings list compensation that includes Restricted Token Units (VVV) alongside salary and stock options, so engineering incentives are tied directly to the token.

Mint vs Buy calculator · Pro ROI · VVV staking / burns path · Related: ASN AS402818

Sources