Blaxel
blaxel.ai ↗Blaxel · YC S25
Blaxel's architecture uses EROFS (read-only base filesystem) + tmpfs (RAM-based writable layer) + OverlayFS orchestration, reserving ~50% of available memory for tmpfs. This enables the 25 ms resume benchmark — the fastest in the industry. Standby transitions complete in ~15 seconds, and the system can sustain unlimited standby duration.
- Total funding
- $7.3M
- Seed · Jul 2025
- Cold start
- <25ms
- Standby resume
- Regions
- 16
- Global
- Team
- ~6
- Co-founders + team
- Founded
- 2024 (core team from ForePaaS / OVHcloud)
- Team size
- 6 co-founders + team
- Funding
- $7.3M Seed (July 2025) led by First Round Capital, with Y Combinator, Liquid 2, Multimodal
- Founders
- Paul Sinaï (CEO) — previously CEO of ForePaaS (acquired by OVHcloud); 5 co-founders all from OVH/ForePaaS
- Customers
- Human Behavior · Vybe · Jazzberry
- Isolation
- Firecracker microVM · EROFS + tmpfs overlay
- Cold start
- Sub-25 ms resume from standby
- GPU
- Yes
- Snapshots
- save+restore+fork — full filesystem and memory; fork from templates
- Storage
- Both — volumes + snapshot-based standby
- Compliance
- SOC 2 Type II · HIPAA BAA · ISO 27001
- GitHub
- Partial OSS (SDK/CLI)
- Last release
- Active development as of April 2026
- Edge
- Industry-fastest resume at 25 ms; perpetual standby with zero compute cost; 16 global regions