Logos identify the maker. Blade AI is an independent supplier.
Live deal · AMD Instinct MI350X
MiTAC 8-GPU servers with AMD Instinct MI350X, available now
Brand-new MiTAC G8825Z5 servers, each with eight AMD Instinct MI350X GPUs (288GB HBM3E each, 2.3TB of GPU memory per server), two AMD EPYC 9575F processors, 3TB of DDR5 and eight 400G network ports, in an 8U chassis. Shipped from the maker about 2 weeks after compliance, with a 5-year warranty confirmed on the quotation. For buyers who need large-memory GPUs for big models now, without waiting for a factory slot. PyTorch, vLLM and SGLang run on AMD ROCm. Sold as complete servers only.
- 8AMD Instinct MI350X GPUs per server, 288GB HBM3E each
- 2.3TBGPU memory per server (an 8-GPU H200 server has 1.13TB)
- 4.6 PFLOPSFP8 per GPU, dense (H200: 1.98)
- About 2 weeksFrom compliance to shipping
The lead-time gap
How fast this one ships
-
AMD Instinct MI350X
MiTAC 8-GPU servers, special offerSpecial offer
Blade AI: ~2 wksThe market: 16–20 wks~16 weeks sooner
Typical lead time from end-user compliance approval. Confirmed on the quotation.
Specification
| System | MiTAC G8825Z5U2BC-350X-575, 8U 8-GPU server |
|---|---|
| GPU | 8 x AMD Instinct MI350X, 288GB HBM3E, 8TB/s each (OAM, on the UBB board) |
| GPU memory, total | 2.3TB per server |
| Processors | 2 x AMD EPYC 9575F (Turin), 64 cores |
| System memory | 24 x 128GB DDR5 RDIMM, 5600MT/s (3TB) |
| Storage | 4 x 1.92TB U.2 NVMe PCIe 4.0, 1 DWPD |
| Network | 8 x 400Gb/s single-port QSFP112 NICs, PCIe 5.0 x16 (Broadcom BCM957608) |
| Expansion | 8 half-height + 4 full-height PCIe 5.0 x16 slots, 2 x M.2 |
| Drive bays | 8 hot-swap 2.5in U.2 NVMe |
| Power | 4+2 hot-swap 13,200W supplies, 80 PLUS Titanium |
| Cooling | 15 hot-swap fans, air-cooled |
| Security | TPM 2.0 |
| Build | Assembled and tested before shipment |
| Warranty | 5 years, confirmed on the quotation |
| Lead time | About 2 weeks after end-user compliance clears |
| Quantity | Limited quantity, subject to prior sale |
| Price | Confirmed on the quotation |
- Special offer: brand new, shipped from the maker about 2 weeks after compliance.
- Sold as complete servers only, not as individual GPUs.
- Offered subject to prior sale and end-user compliance. Specification, quantity and price are confirmed on the quotation.
How it compares
- GPU memory: 288GB per MI350X against 141GB per H200, so 2.3TB per server against 1.13TB.
- FP8: 4.6 PFLOPS per GPU dense against 1.98 on H200; BF16 2.3 PFLOPS; FP4 and FP6 9.2 PFLOPS.
- Bandwidth: 8TB/s per GPU against 4.8TB/s on H200.
- Software: PyTorch, vLLM and SGLang run on AMD ROCm. CUDA still leads for TensorRT-LLM and custom kernels.
H200 or AMD? Read the comparison
Posted 13 November 2026. Specification as supplied. E&OE.
Other live deals
- Supermicro 8-GPU servers with AMD Instinct MI300X, about 6 to 8 weeksAS-8125GS-TNMR2: 8 x AMD Instinct MI300X 192GB, 2 x EPYC 9654, 2.3TB DDR5, 4 x 400GbE, 8U
- NVIDIA A100 GPU racks, complete systems, available nowZT Systems Gen 7.0 48U racks: 32 x A100 40GB SXM4 per rack, from a single rack
- Supermicro 10-GPU servers with AMD Instinct MI350P, available nowAS-5126GS-TNRT2: 10 x AMD Instinct MI350P 144GB, 2 x EPYC 9575F, 2TB DDR5-6400, 5U
Register interest in this deal
MiTAC 8-GPU servers with AMD Instinct MI350X, available now: MiTAC G8825Z5U2BC-350X-575. We reply within one working day with the quantity still open, the price and the lead time.
GPU systems are export-controlled. Before we quote we ask who the end user is and where the hardware will be installed. We do not supply the destinations listed on our export compliance page.
How a live deal works
- Register interest with the quantity, the site and the date you need it.
- We reply within one working day with the specification, the quantity still open and the price.
- Mutual NDA, then end-user and site documents for export screening. We may decline a deal.
- Your purchase order and deposit go to the seller of record. Serial numbers and photos reach you before the balance is due.
