Supermicro, a San Jose server giant, on Monday unveiled a new family of compact platforms for Edge AI — artificial intelligence that operates on-site rather than in the cloud. In stores. In factories. In hospital corridors.
Three new systems built on AMD EPYC 4005 processors with Zen 5 architecture offer varying form factors for different deployment needs. The smallest, AS-E300-14GR, fits into a 2.5-liter case while supporting up to 16 cores and 192 GB of DDR5 memory. The AS-1116R-FN4 provides a half-depth server rack with expanded storage, while the AS-3015TR-i4 uses a 9-liter tower case that accommodates full-size graphics cards such as the NVIDIA RTX PRO 2000 with Blackwell architecture.
Power consumption starts at 65 watts. All three models support PCIe Gen5, TPM 2.0 security modules and remote management through IPMI 2.0. Some configurations include AMD 3D V-Cache technology for accelerated data access.
Mory Lin, Supermicro’s vice president for IoT and Edge Computing, said the new platforms enable deployment of AI accelerators and specialized tasks “at the edge” of networks with lower power consumption, better security and reduced total cost of ownership.
The company positions the systems for specific applications: point-of-sale terminals, theft prevention systems, real-time customer analytics and network gateways. Applications that require instant data processing on-site without sending information to the cloud.
Market dynamics
Edge AI represents one of the fastest-growing segments in the technology market. Analysts estimate the global Edge AI market at approximately $37.5 billion in 2026, with projections reaching $103 billion by 2030. The compound annual growth rate averages about 29%.
The rationale centers on proximity: cloud data centers process data quickly but remotely. When a store security camera must detect theft in milliseconds, transmitting video to servers in another state creates unacceptable delays. Edge AI solves this by performing computations where data originates.
North America accounts for 36% of the global Edge AI market. Silicon Valley, where Supermicro operates, sits at the center of this competition. Rivals including Intel, NVIDIA, Microsoft and Google continue expanding their presence in the segment.
Legal challenges
The product launch comes amid legal troubles for Supermicro.
On March 19, the U.S. Department of Justice charged three individuals connected to Supermicro with smuggling servers containing NVIDIA GPUs to China in violation of export restrictions. Co-founder Yih-Shyan “Wally” Liaw, Taiwan sales manager Ruei-Tsang “Steven” Chang and contractor Ting-Wei “Willy” Sun allegedly orchestrated approximately $2.5 billion in illegal shipments, according to prosecutors. The defendants used shell companies in Southeast Asia, falsified documents and deployed decoy servers during inspections, the indictment alleges.
Supermicro itself faces no charges. The company terminated all three individuals and launched an internal investigation on April 7 led by two independent board directors. The review involves law firm Munger, Tolles & Olson and consulting firm AlixPartners.
Previous controversies include Ernst & Young’s 2024 departure as auditor over internal control concerns. Nasdaq issued delisting warnings before Supermicro filed delayed reports and retained its listing.
SMCI shares gained 8.79% following Monday’s announcement, significantly above the 0.06% average movement on previous AI product launches. The stock has declined 54% over the past six months despite 35% revenue growth.
Founded in 1993 in San Jose, Supermicro maintains headquarters, development and partial manufacturing operations in the city. Additional production occurs in Taiwan and the Netherlands. The company’s market capitalization stands at approximately $15.2 billion.