Observed Signal · Nov 17, 2026 · Technical Release · Source: PR Newswire: Technology News · Impact: 3/5 · Sentiment: Positive
MiTAC Scales Agentic AI with Full-Stack Infrastructure at SC26
At SC26 in Chicago, MiTAC Computing Technology Corporation unveiled a comprehensive full-stack AI infrastructure ecosystem designed to scale agentic AI deployments. The company showcased a disaggregated inference architecture co-developed with Skymizer, which separates compute-intensive prefill (using AMD GPUs) from memory-intensive decoding (using Skymizer HTX301 accelerators), enabling on-premise execution of massive models such as the 2.8-trillion-parameter Kimi K3. MiTAC highlighted collaborations with AMD, Canonical, DDN, Intel, and NVIDIA to deliver end-to-end orchestration, including GPU-as-a-Service. Rack-level innovations include a liquid-cooled 52U rack with 96 AMD Instinct MI355X GPUs achieving 50% higher density, a two-phase direct liquid cooling proof-of-concept with ZutaCore, and high-performance storage via DDN Infinia. Additionally, the Helion Agentic AI Rack, developed with AMD, promises up to 32,768 Zen 6 CPU cores. New server platforms support NVIDIA MGX and AMD EPYC, covering AI, enterprise, and multi-node systems for tailored implementations.
MiTAC's announcement showcases advancements in AI infrastructure, specifically for agentic AI, which is directly relevant to AdTech's increasing reliance on AI for campaign optimization and data processing. The partnerships and new technologies could influence how advertising platforms handle AI workloads, potentially improving efficiency and scalability.
MARKET SIGNALS
Track AMD Signals & Strategic Shifts
Polaris7 connects regulatory filings, primary sources, leadership changes, and M&A activity into structured intelligence. Set up your Polaris7 workspace to monitor relevant companies in your executive briefing.
- Verified primary sources
- Automated briefings
- Instant watchlist setup
Key Takeaways & Evidence Grounding
- MiTAC Computing unveiled a full-stack AI infrastructure ecosystem at SC26 in Chicago for scaling agentic AI workloads.
- MiTAC and Skymizer developed a disaggregated inference architecture using GPU-based prefill servers and Skymizer HTX301 decoding servers.
- The architecture can run the 2.8-trillion-parameter Kimi K3 model on-premise.
- MiTAC's 52U liquid-cooled rack with 96 AMD Instinct MI355X GPUs achieves 50% higher GPU density.
- MiTAC is developing the Helion Agentic AI Rack with AMD, scaling up to 32,768 Zen 6 CPU cores, and collaborated with AMD, Canonical, DDN, Intel, NVIDIA, and Rafay.
Connected Companies & Entities
“Strategic partnerships with AMD, Canonical, DDN, Intel, NVIDIA, and Rafay enable MiTAC to deliver scalable AI factory architectures.”
“Canonical Ubuntu provides the certified operating system layer ensuring smooth deployment and reliability.”
“Strategic partnerships with Intel, among others, support MiTAC's full-stack infrastructure for AI.”
“MiTAC offers servers with NVIDIA MGX architecture and NVIDIA RTX PRO Blackwell GPUs.”
Ontology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
MiTAC Computing Showcases Full-Stack AI Infrastructure at SC26
MiTAC Computing Technology Corporation, a subsidiary of MiTAC Holdings, unveiled its comprehensive AI infrastructure ecosystem at the SC26 conference in Chicago. Under the theme 'Scaling agent-based AI. Built on full-stack infrastructure,' the company demonstrates how innovations across the technology stack overcome bottlenecks in compute capacity, thermal management, and data for enterprise AI deployments. Key highlights include a disaggregated AI inference engine developed with Skymizer, featuring the MiTAC G8825Z5 prefill server with AMD EPYC and Instinct GPUs, and the G4520G6 decode server with Skymizer HTX301 accelerator cards. MiTAC also showcased a Helion Agentic AI Rack in collaboration with AMD, high-density liquid-cooled racks with 50% greater GPU density, and a storage rack powered by DDN Infinia. The ecosystem integrates software from Canonical Ubuntu and Rafay, with orchestration via MiCoreView.
AMD acquires Taalas to hardwire AI into silicon
Advanced Micro Devices (AMD) has agreed to acquire Taalas, a Toronto startup that builds inference accelerators customized—or hard-wired—for a single AI model. Taalas says its specialized chips can produce outputs for specific models thousands of times faster than traditional GPUs in exchange for less flexibility. The company, founded in 2023, has raised $219 million; its current chip runs a small version of Meta's Llama 3.1 and is manufactured on an older TSMC process using on‑chip SRAM. AMD plans to integrate Taalas' technology into its roadmap and systems alongside its CPUs and Instinct GPUs, including rack-scale offerings such as Helios. The acquisition follows a wave of AI-hardware deals, including Nvidia’s purchase of Groq assets for $20 billion seven months earlier, and fits AMD’s broader buying strategy to assemble components for integrated AI systems.
Microsoft Unveils Maia 200 Inference Accelerator
ChipStrat interviewed Saurabh Dighe, CVP of Azure Systems and Architecture at Microsoft, about Maia 200 — Microsoft’s second-generation AI accelerator designed specifically to optimize inference economics. Maia 200 targets performance-per-dollar and performance-per-watt for inference, with architectural trade-offs that favor inference workloads over training. Key technical choices discussed include a much larger on-die SRAM, a memory hierarchy balancing SRAM, HBM and system DRAM, and a preference for a large Ethernet-based scale-up domain with a custom transport layer. Microsoft positions Maia 200 as complementary to merchant GPUs within a heterogeneous fleet, exposing capacity through Azure services rather than as a standalone product. The interview also emphasizes KV cache management for long-context workloads and the importance of software investments (compilers, kernels, pre-silicon tooling) ahead of silicon.
WATCHLIST & BRIEFINGS
Receive future signals for AMD & competitors in your briefing
Stay ahead of material M&A deals, leadership changes, account wins, and product rollouts. Polaris7 connects primary filings, company newsrooms, and trade reporting directly into your executive briefing.
- Regular primary-source checks
- Structured Watchlist briefings
- Preserves current context
Market Signal Monitoring
Follow affected companies in Polaris7 with regular source checks and structured Watchlist briefings.
Monitor AMD
Follow affected companies in Polaris7 with regular source checks and structured Watchlist briefings.
