GPU + CPU cloud · built for builders

Run heavy
workloads. From a penny.

Dedicated NVIDIA GPUs from $0.15/hr. Lean CPU instances from $0.01/hr. Free 250GB backups, browser console, per‑VM firewall, and BYOK Windows — included on every VM.

0 egress fees. Move data freely·multiple 10G uplinks, 1Gbps per VM, no metering.
$0.01
CPU / hr
250GB
Free backups
30s
Deploy time
99.9%
Uptime SLA
Pick a VM
Live pricing
Type Cores RAM Disk CPU Price
v1_cpu_nano 1 2GB 50GB Intel Haswell / Scalable $0.01/hr
v1_cpu_small 1 8GB 150GB Intel Haswell / Scalable $0.03/hr
v1_cpu_medium 4 16GB 250GB Intel Haswell / Scalable $0.06/hr
v1_gpu_1x_2080ti 4 16GB 250GB Intel Haswell $0.15/hr
v1_gpu_1x_a4000 4 16GB 250GB Intel Scalable G2 $0.20/hr
v1_gpu_1x_a5000 4 32GB 250GB Intel Scalable G2 $0.30/hr
v1_gpu_1x_a6000 8 32GB 250GB Intel Scalable G2 $0.55/hr
v2_gpu_1x_5090 16 64GB 500GB AMD Milan $0.75/hr
From $0.01/hr
Deploy

Two ways to spin up.
One that costs less than coffee.

Pay by the hour, billed by the second. No commits, no minimums, no hidden egress fees. Bring your workload, leave when you want.
New · CPU

CPU instances from a penny.

Dedicated cores for jump hosts, build agents, light services, and side projects. No GPU tax — just clean, fast compute.

$0.01/ hr
1 dedicated core
2 GB RAM
50 GB SSD
root + ssh
Spin up a CPU box
Most popular

GPU instances from $0.15/hr.

RTX 2080 Ti at $0.15/hr on Intel Haswell hosts. Step up to A4000 / A5000 / A6000 on Intel Xeon Scalable Gen 2, or RTX 5090 on AMD Milan. No oversubscription, drivers ready.

$0.15/ hr · 1× RTX 2080 Ti
4 dedicated cores
16 GB RAM
250 GB SSD
Intel Haswell
Pick a GPU

Everything you'd build yourself,
already built.

Console, backups, firewall, MFA, BYOK Windows, extra IPs — the operational scaffolding around your VM is included by default.
Browser console

SSH locked you out? Console saves you.

A live VNC stream straight to the VM — even when networking is misconfigured, the firewall is wrong, or you broke sshd. Send keys, full‑screen, scroll back. No agents installed.

Send Keys
Fullscreen
Power menu
Browser-based VNC console
Power

Reboot, shutdown, full power cycle.

Wedged kernel? Stuck installer? One click for a hard power cycle — exactly like a PDU button, no ticket required.

Power controls dropdown
Firewall

Per‑VM firewall without the docs.

Add ACCEPT or DROP rules in seconds. Auto‑detects your current public IP for "lock it down to me" SSH rules.

Add firewall rule modal
Account security

MFA on every account, not just admins.

TOTP via Google Authenticator, 1Password, or any app you already trust. Backup codes, account‑lockout protection.

Two-factor authentication setup
Backups

Snapshot anywhere, 250GB free.

Schedule recurring backups or hit "Create Backup" before a risky migration. Restore back to any point.

250GB
Included on every account
0used 80GB250GB
Operating systems

Windows 10 / 11 — BYOK.

Bring your own license, boot a Windows VM in seconds. Or use Ubuntu, Debian, or upload a custom ISO.

Bring Your Own Key retail · OEM · volume license
Networking

Need more IPs? $3/mo each.

Every instance ships with a public IPv4. Add more for SNI TLS, mailservers, or anything that needs its own address.

203.0.113.42 Primary
203.0.113.43 +$3/mo
  Add another IP +$3.00/mo

IGL runs production.
PHL is the safety net.

Two locations, one platform. Same API, same image library, same backups across both — only the iron differs.
Production
igl · flagship

Wilmington, DE

Our flagship Tier‑3 datacenter — DataPath, 23,000 sqft of concrete-vault facility, multi-carrier fiber ring. This is where the platform actually runs.

  • Tier‑3 datacenter · concrete vault construction
  • 10Gbps multi-carrier uplinks · redundant fiber
  • Full GPU lineup — 2080 Ti, A4000, A5000, A6000, 5090
  • SSAE‑18 audited · biometric access
Deploy in IGL
Standby
phl · backup

Philadelphia, PA

Our backup site — a small corporate network in our Philadelphia office, not a real datacenter. Useful as a failover target for DR; limited capacity by design.

  • Best for failover & DR — not for primary deploys
  • 1Gbps uplink · single-carrier office network
  • ~20× RTX 2080 Ti hosts plus legacy capacity
  • Same API · cross-region restore from IGL
Use PHL for redundancy

From SolidWorks.
To a GPU cloud.

Short version: I'm a telco operator. When COVID hit in 2020 I started Alpinevans.com — building Sprinter vans, designing them in SolidWorks on workstations stacked with NVIDIA cards. When I wound the van business down in 2023, those RTX 8000s and 3090s were just sitting there. I put them on vast.ai, then TensorDock — telco instincts said both could be run a lot better, so I built Blue Lobster. Today AS398235 is a registered ASN, real BGP from IGL, hosting our own products (Concentrate.ai) on the same iron you'll deploy on.
2020

Alpinevans

COVID hit. Started Alpinevans.com — custom Sprinter builds, SolidWorks CAD on workstations stacked with NVIDIA cards.

2023

Shut down · vast.ai

Wound the van business down. The RTX 8000s went onto vast.ai. First taste of GPU rental — and what those marketplaces were missing.

2024

Blue Lobster · TensorDock

Blue Lobster is born. 3090s up on TensorDock. Quietly started building a real platform — telco-grade ops, dedicated hardware, real backups.

2025

bluelobster.ai

Public launch. Registered AS398235, real BGP at IGL, hosting our own products (Concentrate.ai) on this exact platform.

80%
Cheaper than hyperscalers
No marketing math. Match a g4dn or g5 spec, you'll see.
30s
VM provisioning
Click deploy → SSH in. We don't queue you behind anyone.
2 regions
PHL · IGL
More on the way. East coast and Midwest US datacenters.
0 egress fees
Move data freely
Multiple 10G uplinks, 1Gbps per VM, no metering.
· Start cheap. Scale when you're ready.

Ship something today
for a penny an hour.

Free to sign up. No credit card to browse. Start a CPU box for $0.01/hr or a GPU box for $0.15/hr — billed by the second.