Key Takeaways
- ASUS has broadened its AI‑infrastructure portfolio through new alliances with Intel, AMD and a dozen‑plus technology partners.
- Two Intel‑centric rack servers (RS700‑E12‑RS4U, RS720‑E12‑RS12U) and the high‑density XA P8I‑E13A deliver up to six‑socket Xeon Scalable CPUs and native GPU acceleration.
- AMD‑driven platforms (RS720A‑E14B‑R32U, RS500A‑E14B‑R12U, ESC8000A‑E13P) pair 3rd‑gen EPYC 9006 CPUs with Instinct MI350P accelerators for maximum throughput.
- A side‑by‑side spec table shows where Intel and AMD solutions diverge in core count, memory bandwidth, and power envelope.
- The expanded ecosystem is positioned to accelerate AI adoption in enterprise, HPC and edge workloads while delivering a more resilient, governed “AI factory” architecture.
Overview of ASUS’s AI‑Infrastructure Expansion
At the ASUS AI Tech 2026 showcase, the Taiwanese OEM announced a strategic push to turn its server lineup into a full‑stack AI “factory.” By locking‑in design‑win agreements with Intel, AMD, and a broader partner network—including Aleria, Foxlink, IBM, Samsung, Schneider Electric, and Western Digital—ASUS aims to cover every layer of the AI stack, from silicon to storage and power management.
The announcement underscores three core objectives:
- Performance density – packing more compute and memory into smaller rack footprints.
- Resilience & governance – leveraging partner‑provided monitoring, security and power‑optimisation tools.
- Scalability – enabling seamless scale‑out from a single 4U chassis to a 12U data‑center pod.
Partner Network That Powers the AI Factory
| Partner | Primary Contribution | Example Integration |
|---|---|---|
| Intel | Xeon Scalable CPUs, Optane persistent memory | RS700‑E12‑RS4U, XA P8I‑E13A |
| AMD | EPYC 9006 CPUs, Instinct MI350P GPUs | ESC8000A‑E13P |
| IBM | AI‑optimized firmware and Red Hat OpenShift support | Container orchestration |
| Samsung | High‑capacity NVMe SSDs (up to 30 TB) | Storage tiering |
| Schneider Electric | Intelligent PDUs & power‑capping APIs | Energy‑aware scheduling |
| Western Digital | Enterprise‑grade HDD/SSD arrays with AI‑ready data reduction | Tiered cold‑storage |
These collaborations give ASUS customers a “plug‑and‑play” path to build AI‑ready clusters without the typical integration overhead.
Intel‑Powered AI Infrastructure
Core Server Models
| Model | Form Factor | CPU Configuration | GPU Support | Memory Capacity | Typical Use‑Case |
|---|---|---|---|---|---|
| RS700‑E12‑RS4U | 4U rack | Up to 6‑socket Intel Xeon Scalable (up to 40 cores per socket) | 4× PCIe 4.0 x16 slots (NVIDIA A100, AMD Instinct) | 4 TB DDR5 (3200 MT/s) | Edge AI & low‑latency inference |
| RS720‑E12‑RS12U | 12U rack | Same CPU options, plus optional Intel Optane DC PMem 1.5 TB | 8× GPU bays (up to 8 × A100) | 8 TB DDR5 | Large‑scale training & HPC |
| XA P8I‑E13A | 6U high‑density | Single‑socket Intel Xeon Next‑Gen (up to 64 cores) | Integrated GPU‑direct lanes for up to 4 × NVIDIA H100 | 2 TB DDR5 + 2 TB Optane | AI inference at the network edge |
All three platforms ship with Intel vPro remote management, IPMI 2.0 and support for Intel® oneAPI toolkits, enabling developers to compile a single code base for CPUs, GPUs and FPGAs.
AMD‑Powered AI Infrastructure
Core Server Models
| Model | Form Factor | CPU Configuration | GPU Accelerator | Memory Capacity | Distinctive Feature |
|---|---|---|---|---|---|
| RS720A‑E14B‑R32U | 32U rack (modular) | Up to 4‑socket AMD EPYC 9006 (up to 96 cores total) | 4 |