The problem
Serving AI models on GPUs is costly and power-hungry, and scarce HBM memory caps supply
What they do
Memory-first AI inference servers and custom chips sold to clouds and AI providers
Traction 50+ Atlas racks at Oracle Cloud; Jump Trading, Parasail, i3d.net customers; ~$5.4M 2025 revenue (est.)
Hiring trend
- Open roles
- 30
- Ops roles
- 1
- Headcount
- ~76-98
Headcount 49 (Dec 2025, LinkedIn via Latka) → ~98 (Tracxn, 2026), roughly doubled; 18 open roles, mostly ASIC/silicon in Austin and remote US
Glassdoor
No rating yet · see Glassdoor ↗Funding
- Last round
- $875M
- Stage
- Series C
- Date
- Sep 2026
Led by NEA, Atreides Management, Valor Equity Partners, Andra Capital, SemiAnalysis Capital, Jim Clark
- Hiring+230 open roles on the live job board (30+ roles = +2)
- Funding recency+2Series C · $875M, Sep 2026 (1 month ago): 6 months or less = +1.5, $100M+ round +0.5
- Traction+0.75Some growth: 50+ Atlas racks at Oracle Cloud; Jump Trading, Parasail, i3d.net customers; ~$5.4M 2025 revenue (est.)
- Red flags0None found in the last 24 months
- Total4.752 + 2 + 0.75 − 0 = 4.75 → Sprint (4.75 or more)
Distance (10K) is a separate read on how long the lead can last. It doesn't change the score.
Distance
10K · edge: cheaper at size
Real Oracle deployment and a differentiated HBM-free bet, but a follower behind Nvidia and better-established inference-chip rivals, with next-gen silicon not shipping until H2 2027 and heavy burn.
Sources: PR Newswire: Positron $875M Series C at $5B (Sep 2026) ↗ · GetLatka: Positron AI revenue estimate ↗
Open ops roles · 1
Last checked Oct 25, 2026