Super Micro Computer shares fell 13.5% in December after a year-long decline driven by investor nervousness over an AI datacenter slowdown. The company, which saw revenue rise to $21 billion over the trailing twelve months, warned that a delay in new NVIDIA chip shipments and a potential oversupply of AI accelerators could push projected fiscal-2026 revenue of $36 billion below expectations. Gross margins, which currently hover around 10.1%, leave little buffer for profit if end-customer cloud buildouts soften. Net income of just under $800 million over the same period translates to a trailing P/E of 24, raising questions about valuation in the event of a chip spending downcycle.

On January 6, Super Micro Computer entered into a senior revolving credit agreement with JPMorgan Chase Bank as administrative agent and a syndicate of lenders. The facility commits $2.0 billion through December 29, 2030 and may be used for working capital, letters of credit and general corporate purposes. Obligations under the agreement are secured and include customary covenants and default triggers. CEO Charles Liang noted the facility will support both day-to-day operations and planned production ramp-ups, preserving liquidity as the company accelerates its AI-focused initiatives.

Super Micro Computer announced expanded production and direct liquid cooling capabilities for NVIDIA’s next-generation Vera Rubin NVL72 SuperCluster and HGX Rubin NVL8 systems. The NVL72 SuperCluster leverages 72 Rubin GPUs and 36 Vera CPUs to deliver 3.6 exaflops NVFP4 and 1.4 PB/s HBM4 bandwidth, while the 2U HGX Rubin NVL8 offers 400 petaflops NVFP4 in an enterprise-optimized form factor. By integrating in-row coolant distribution units and modular Data Center Building Block Solutions, the company projects a 30% reduction in energy and water consumption per rack and a 25% faster deployment cycle versus traditional air-cooled designs.