Where AI actually runs
Most of every AI pipeline is CPU work: ingesting and cleaning data, orchestrating embeddings, retrieval, tool calls, agent loops and the APIs that serve results.
Platform · CPU Compute
CPU compute, the backbone of AI.
AI, data and memory run on CPUs first. Data preparation, retrieval, vector search, orchestration, agents, caches and serving all live here, while GPUs accelerate the heavy math. Launch general purpose, compute optimized and memory optimized instances across 13 regions in minutes.
$600 in launch credit + 20M free inference tokens. No commitment.
Engineered for
- 13 global regions
- CPU & GPU compute
- Self-healing orchestration
- Zero-trust security
- OpenAI-compatible API
- Sovereign & air-gapped options
CPU Compute
Why teams choose it.
General purpose, compute optimized and memory optimized CPU instances, the backbone of every AI pipeline.
Memory where it matters
Memory optimized instances with up to 256 GB of RAM keep vector indexes, feature stores, caches and in-memory analytics hot, so answers come back in milliseconds.
Right-sized, always
Three families and 9 instance shapes let you match every service to the right balance of cores and memory, and pay only for what you need.
Capabilities
Every CPU shape an AI stack needs.
Every capability is available through the dashboard, the API and the AI ops copilot.
General purpose
Balanced vCPU and memory for APIs, web backends, orchestration services, schedulers and development environments.
Compute optimized
Dedicated vCPUs, up to 32 per instance, for tokenization, ETL, batch pre- and post-processing, encoding and high-throughput request handling.
Memory optimized
Up to 256 GB of RAM on dedicated vCPUs for vector databases, retrieval indexes, caches, graph workloads and in-memory analytics.
Retrieval and vector search
Run your vector database and search tier next to your models, on the same private fabric, with predictable latency.
Prepaid and predictable
Choose an instance type, a count and the hours you need, then pay upfront or from your compute balance. Unused time returns to your balance.
Self-healing by default
Unhealthy hosts are drained and replaced automatically, and your services come back on healthy capacity without anyone filing a ticket.
- Instance families
- 3
- Max vCPUs per instance
- 32
- Max memory per instance
- 256 GB
- Global regions
- 13
Figures describe platform capabilities and service-level targets. Contracted commitments are set out in our Service Level Agreement.
Global cloud fabric
CPU Compute, everywhere you operate.
Available across 13 regions on six continents, connected by a private, encrypted fabric with automatic failover and per-region data residency.
- US East · Ashburn
- US West · Hillsboro
- Canada Central · Montréal
- South America · São Paulo
- EU West · Dublin
- EU Central · Frankfurt
- EU North · Stockholm
- Middle East · Dubai
- Africa South · Johannesburg
- Asia South · Mumbai
- Asia Southeast · Singapore
- Asia Northeast · Tokyo
- Australia East · Sydney
Use cases
What teams build with CPU Compute.
Starting points that deliver value in days.
Data preparation
Ingest, clean, deduplicate and tokenize datasets before a single GPU spins up.
Retrieval and RAG
Host vector indexes and retrieval services that ground every answer in your own data.
Agents and orchestration
Run agent loops, tool calls, queues and workflow engines that coordinate models end to end.
In-memory analytics and caches
Keep hot data, feature stores and session memory in RAM for millisecond responses.
Inference pre- and post-processing
Handle tokenization, routing, guardrails and formatting around every model call.
APIs and serving
Serve the applications, gateways and dashboards your users touch every day.
Security & compliance
Secure at every layer.
Zero-trust access, encryption everywhere and immutable audit trails are standard on every product, on every plan.
- Isolated networking, encrypted volumes and dedicated vCPUs on compute and memory optimized instances.
- Encryption at rest and in transit, with customer-managed keys available.
- SSO, role-based access control and scoped API keys.
- Your data is never used to train shared models.
- Built to align with SOC 2, ISO 27001, HIPAA and FedRAMP-moderate control frameworks.
Zero-trust by default
Every request authenticated, authorised and encrypted.
Built to align with control frameworks
- SOC 2
- ISO 27001
- HIPAA
- GDPR
- FedRAMP Moderate
- PCI DSS
AES-256
at rest
TLS 1.3
in transit
Immutable
audit trail
FAQ
Frequently asked questions
Straight answers from the team that runs the fabric. Still curious? Our architects reply fast.
Why does AI need CPUs?
Because most of an AI system is not matrix math. Data preparation, retrieval, vector search, orchestration, agents, memory-heavy caches and the APIs that serve results all run on CPUs. GPUs accelerate training and heavy inference; CPUs carry everything around them.
Which CPU instances are available?
Three families: general purpose, compute optimized with dedicated vCPUs, and memory optimized with up to 256 GB of RAM. Every instance type and its hourly rate is listed on the Pricing page.
How do I pay for CPU compute?
CPU time is prepaid. Pick an instance type, how many instances and how many hours, then pay upfront or from your compute balance. Paid time only counts while your instances are running, unused time returns to your compute balance, and optional auto-renew keeps long-running services online.
Can I use CPU and GPU compute together?
Yes. CPU and GPU compute share the same private fabric, console, API and compute balance, so your data, retrieval and orchestration tiers sit right next to your accelerators.
Build your AI stack on the right foundation.
Start with $600 in launch credit and 20M free inference tokens, double the typical big-cloud free trial. No commitment, live in minutes.
Double the typical big-cloud free trial · Per-second metering · Live in minutes