Bitdeer AI2026-10-09 12:29:45Bitdeer AI Secures 67MW Data Center Campus in Malaysia for Nvidia GB300 SystemsBitdeer AI has secured a 67MW data center campus in Malaysia and plans to use the site to host Nvidia GB300 systems, according to Odaily. Following the expansion, the company said its secured capacity for AI cloud operations has risen to 273.5MW. It also said the scale of its reserve project pipeline for the AI cloud business is expected to exceed $10 billion. The update points to a larger infrastructure buildout tied to Bitdeer AI’s cloud computing plans, with the newly secured Malaysia campus set to support deployments based on Nvidia’s GB300 platform. No additional timeline or project details were disclosed in the brief.20
Bitdeer AI2026-10-09 12:26:04Bitdeer AI secures 67MW data center campus in Malaysia for Nvidia GB300 systemsBitdeer AI said on Oct. 9 that it has secured a 67MW data center campus in Malaysia, with the site planned to support Nvidia GB300 systems. Following the expansion, the company said its secured AI cloud capacity has increased to 273.5MW. It also estimated that its reserve project pipeline for AI cloud services now exceeds $10 billion. The update centers on infrastructure capacity and project pipeline scale, with the company framing the Malaysia site as part of its AI cloud buildout.20
OpenAI2026-08-25 21:10:46OpenAI's Jalapeno Chip Tops Nvidia GB300 in Power Efficiency and Response SpeedTechub News, citing Bloomberg, says OpenAI’s in-house Jalapeno processor outperforms Nvidia’s GB300 in power efficiency and response speed. The report says the result marks a major step for OpenAI in AI chips. It also says the company’s self-developed hardware could lower model inference costs and reduce reliance on third-party chips.910
Gigabyte2026-08-21 09:59:37Gigabyte launches W775 desktop AI supercomputer powered by NVIDIA GB300 UltraGigabyte said on Aug. 20 that it has launched the W775 series desktop AI supercomputer, with the flagship model using NVIDIA’s GB300 Grace Blackwell Ultra chip. The company says the system brings data-center-grade computing into an office-ready floor-standing chassis and extends its AI lineup from data centers to edge and enterprise deployments. The W775-V10-L01 can support up to 400 simultaneous users and delivers peak throughput of 9,000 tok/s. Gigabyte also raised its full-year capital expenditure forecast to NT$28 billion to NT$30 billion, up from NT$23 billion to NT$25 billion, an increase of about 20%.1340
Nvidia2026-07-27 00:41:14Nvidia’s first U.S.-made GB300 AI chips roll off TSMC’s Arizona lineBlockBeats reported on July 27 that Nvidia’s first U.S.-made GB300 AI chips have rolled off the production line at Taiwan Semiconductor Manufacturing Co.’s Arizona plant. Media reports said CoWoS advanced packaging will be the final step needed to make the GB300 fully localized in the United States. The chip is built on a 4nm process and is also referred to as Grace Blackwell Ultra. The update points to progress in bringing a key Nvidia AI product into domestic manufacturing, with packaging still identified as the last remaining part of the localization process.1920
Qingyun2026-07-23 02:40:15Qingyun Replaces CEO After Nvidia GB300 Smuggling Probe, Stock Rebounds After Three Limit-Down SessionsQingyun replaced its general manager after being drawn into a probe tied to alleged Nvidia GB300 server smuggling to China. The stock had fallen from NT$524 to NT$382.5 in three sessions before rebounding to limit-up.580
Foxconn2026-07-20 03:17:02Foxconn Wins Reported $52 Billion SpaceX Order for NVIDIA GB300 AI Server RacksFoxconn has reportedly secured SpaceX’s first batch order for AI server racks built around NVIDIA’s GB300 platform, in a deal that Economic Daily News said could be worth $52 billion. The report says the order covers roughly 13,000 racks at about $4 million each, representing around 936,000 NVIDIA GPUs, with shipments expected from the fourth quarter of 2026 through the first quarter of 2027. Foxconn did not comment on specific customers or order developments, but said its AI server business has strong order momentum and better-than-expected visibility, while maintaining an optimistic full-year outlook. The reported win would break a supply pattern long dominated by Dell and Supermicro in SpaceX’s AI server outsourcing, and would give Foxconn a direct relationship with what the report described as the last missing major cloud-service customer in its portfolio. The article also linked the order to broader demand from SpaceX’s computing expansion, including multi-year compute rental agreements with Anthropic and Google, plus plans centered on AI1 satellites using NVIDIA GB300 racks. ABMedia further highlighted Foxconn’s exposure to CPO switching through Fii, along with recent earnings figures for both Foxconn and Fii.2220
Nvidia2026-07-06 03:29:56Nvidia GB300 NVL72 Hits 61,400 Concurrent AI Agents per Megawatt in AA-AgentPerf BenchmarkNvidia said its GB300 NVL72 system delivered 61,400 concurrent AI agents per megawatt in the new AA-AgentPerf benchmark from independent evaluator Artificial Analysis, compared with roughly 2,600 for the previous-generation H200. Under the benchmark’s 20-token-per-second and 60-token-per-second service tiers, Nvidia said the GB300 maintained about a 20x advantage on the key “agents per megawatt” metric, while per-GPU density reached 57.5 agents versus 1.4 on H200, implying about a 40x gap on that measure. The significance of the result is not only the hardware gain, but also the benchmark’s design. AA-AgentPerf is positioned as a benchmark built specifically for AI agents rather than single-shot inference. It replays real coding-agent trajectories, includes long context windows, tool interactions, and chained model calls, and evaluates systems under service-level objectives such as output speed and time-to-first-token. The shift suggests AI inference evaluation is moving away from token throughput alone toward system-level efficiency, service density, and the number of agents a power budget can sustain.500
