QumulusAI orders 1,632 Nvidia Blackwell B300 GPUs to expand AI cloud infrastructure

Danucha Pichayanan, Secretary-General of the National Economic and Social Development Council, in his capacity as a board member and secretary of the committee on policy for the operation of data centre businesses, disclosed that the criteria governing the data centre business are expected to be finalised and begin actual enforcement no later than mid-October, after various agencies submitted their data, covering current service providers, criteria for classifying data centre sizes, and average electricity and water usage. On the electricity criteria, the Ministry of Energy will have to set criteria controlling consumption volumes, with electricity rates to apply specifically to the data centre business group. As for the criteria governing each category of size, operators that meet the industrial-type criteria must be located in designated industrial estate areas, and this must not affect the household sector. He also expressed the view that hyperscale data centres should be set at an electricity usage level of 100 megawatts or more. The clarity of the actual criteria must await the board's consideration. On the timeline, after the deadline, a conclusion is to be reached within one month from the first meeting on 4 September, after which the process of drafting a policy specifically for data centres will begin, and a mechanism must be considered to prevent operations from becoming fragmented. As for which agency will be the host, that must await a further conclusion.

NVIDIA, Google, and Emerald AI have created the AI Energy Management Alliance to coordinate electricity use in AI data centers, and the group has started building what it calls the first purpose-built power-flexible AI data center, designed to adjust consumption with grid needs. AEMA plans to publish shared frameworks for grid-responsive data centers with input from utilities, regulators, and large AI infrastructure users. The alliance's first power-flexible facility is in Manassas, and the inflection point will be whether it secures faster grid interconnection terms or preferential tariffs by late 2026. Concrete regulatory wins or standardized grid-responsive design templates would show AEMA is more than marketing and could influence how future DSX style AI factories are specified and approved. NVIDIA operates as a data center scale AI infrastructure provider across the US, Asia, Europe, and other regions, so any shift toward power-flexible facilities directly affects how its hardware and software fit into future grid-aware computing projects.