Intel Xeon dedicated servers. When Intel is the requirement.
For software certified on Intel platforms, and for AI inference that can use AMX: a 24-core Xeon Gold 6312U, and a 64-core Xeon 6761P with 136 PCIe 5.0 lanes and eight channels of DDR5-6400.
- 24–64cores
- AMXmatrix engines on Xeon 6
- 136PCIe 5.0 lanes (6761P)
- $179from, per month
Which one for which job?2 in stock
- Intel Xeon Gold 6312UIntel-certified applications and AVX-512 code on 24 Ice Lake cores, from $179.$179/mo
- Intel Xeon 6761PThe flagship: 64 Granite Rapids cores, AMX, QAT and DDR5-6400 for inference and heavy compute.$469/mo
Somewhere in between? Ask an engineer — median first reply under 15 minutes.
Intel Xeon available now
Monthly prices in USD with setup, bandwidth, DDoS filtering and IPMI included. Switch the billing cycle to see the effective monthly price.
| Network | Locations | Order | ||||
|---|---|---|---|---|---|---|
|
|
128 GB DDR4 ECC8 × 16 GB · RDIMM 3200 | 2 × 1.92 TB NVMeGen4 datacenter · RAID 1 | 1 Gbps unmetered |    |
Deploy Intel Xeon Gold 6312U | |
|
|
256 GB DDR5 ECC8 × 32 GB · RDIMM 6400 | 2 × 1.92 TB NVMeGen4 datacenter · RAID 1 | 1 Gbps unmetered |  |
Deploy Intel Xeon 6761P |
No server matches these filters right now.
- $0 setup fee
- IPMI / KVM console
- DDoS protection
- 1 IPv4 + /64 IPv6
- Free hardware replacement
- Cancel anytime
Compare the Intel Xeon range
Manufacturer specifications and the exact configuration we deliver.
| Specification | Intel Xeon Gold 6312U | Intel Xeon 6761P |
|---|---|---|
| Architecture | Ice Lake-SP | Granite Rapids (Xeon 6700P) |
| Cores / threads | 24 / 48 | 64 / 128 |
| Base / boost | 2.4 GHz / 3.6 GHz | 2.5 GHz / 3.9 GHz |
| L3 cache | 36 MB | 336 MB |
| Memory | 128 GB DDR4 ECCDDR4-3200, 8 channels | 256 GB DDR5 ECCDDR5-6400, 8 channels |
| Storage | 2 × 1.92 TB NVMeGen4 datacenter · RAID 1 | 2 × 1.92 TB NVMeGen4 datacenter · RAID 1 |
| PCIe | PCIe 4.0, 64 lanes | PCIe 5.0, 136 lanes |
| Vector ISA | AVX-512, 2 FMA units | AVX-512 + AMX (INT8, BF16, FP16), QAT |
| TDP | 185 W | 350 W |
| Network | 1 Gbps unmetered | 1 Gbps unmetered |
| Price | $179/mo | $469/mo |
What you get with this range
AMX for inferenceAdvanced Matrix Extensions accelerate INT8, BF16 and FP16 inference in PyTorch (oneDNN) and OpenVINO — no GPU required.
Vendor certificationSome commercial software is certified or licensed for Intel platforms only. These servers keep you inside the support matrix.
AVX-512 with two FMA unitsBoth CPUs execute AVX-512 natively with two FMA units per core — strong for scientific and media code built for Intel.
QuickAssist on Xeon 6The 6761P integrates QAT to offload compression and encryption, freeing cores for your workload.
Intel Xeon questions
Ask the engineers who run the servers — median first reply under 15 minutes, 24/7.
Contact usWhat is AMX, and does my software use it?
AMX is a set of matrix-multiplication units in each Xeon 6 core. Frameworks built on oneDNN (PyTorch, TensorFlow) and OpenVINO use it automatically for BF16 and INT8 work; plain C code does not.
Is QuickAssist (QAT) available to me?
Yes, on the Xeon 6761P. With the QAT driver installed, compatible software — OpenSSL engines, zlib/zstd acceleration — offloads work to it.
Why only a few locations?
Xeon servers are built in smaller batches: the 6312U is stocked in London, Frankfurt and Miami, the 6761P in Amsterdam and Frankfurt. Other cities are possible as a custom build.
Can I upgrade the memory?
Yes: the 6312U to 256 GB and the 6761P to 512 GB, installed in place.