Supermicro lands a $60 billion AI server backlog
Supermicro told investors in a preliminary earnings forecast that it has secured more than $60 billion in new orders for its fiscal fourth quarter, sending its shares up more than 19% in extended trading. The company now expects fourth-quarter gross margins of 15% to 17%, well above its prior forecast of 8.2% to 8.4%, citing a favorable customer and product mix. Backlog reached "record levels" at the end of fiscal 2026, though revenue is expected to come in at the lower end of the previously guided $11 billion to $12.5 billion range, with full results due on August 11.
SpaceX leads a new class of neocloud buyers
DigiTimes reports that SpaceX is among the neocloud customers driving Supermicro's order book, reshaping demand toward rack-scale AI infrastructure rather than discrete boxes. The relationship is highlighted as evidence that rack-scale systems are beginning to translate into profitability, not just volume. Supermicro has been working to speed up fulfillment for advanced AI servers from more than 20 customers, partly financed by a $7 billion equity and equity-linked offering announced earlier.
Nvidia's Vera Rubin targets the AI factory CPU bottleneck
Nvidia launched the Vera Rubin platform, a seven-chip full-stack system designed around the demands of agentic AI. The platform features a custom Arm-based processor built on Nvidia's new Olympus core, with 88 custom cores, 176 hardware threads, up to 1.2 terabits per second of LPDDR5X memory bandwidth, 164 megabytes of unified L3 cache, and up to 1.8 TB/s of coherent CPU-GPU bandwidth via NVLink-C2C. Nvidia argues that traditional chiplet-heavy server CPUs leave performance on the table for latency-sensitive agentic workloads, framing Vera as a response to what it calls the "max single-threaded CPU at scale."
Hyperscalers begin deploying Vera Rubin
Nvidia says Vera Rubin is already being deployed by partners including CoreWeave, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure. The rollout positions the new platform against incumbent server CPU vendors and extends Nvidia's reach deeper into the data center stack. Suppliers such as Supermicro, Dell Technologies, and Hewlett Packard Enterprise have benefited from surging demand for AI server hardware as cloud providers and enterprises scale up large language model infrastructure.
Share this article
