In its 2026 sustainability report, Meta acknowledged it is re-evaluating the 2030 timeline and interim milestones for its net-zero emissions goal, citing AI infrastructure demand that has grown faster than almost anyone projected. The company posted a record percentage increase in greenhouse gas emissions, outpacing every other Big Tech peer, while insisting its long-term ambitions remain unchanged.
AI Data Center
AI 数据中心
High-power-density compute facilities for LLM training and inference, with far higher energy and emissions intensity than tradit
Curated Intelligence
President Trump said at a White House briefing that US companies building AI data centers abroad would face "a tremendous tax," likening it to semiconductor offshoring. He also urged domestic communities to stop rejecting projects and charge developers instead, naming West Virginia as an alternative to Texas or California. No concrete policy was detailed, but the threat could raise cost expectations for firms seeking cheaper power overseas.
Fiber-optic drones have become a staple of the Ukraine battlefield because they resist electronic jamming, but the same fiber supply is now contested by global AI data center demand, driving up import costs. Ukrainian drone makers are adapting by building their own spooling and production capacity. The case shows AI infrastructure demand spilling over into defense supply chains.
SoftBank Group posted 4.24 trillion yen in revenue for the first half of FY2027, up 13.3% year over year, but swung to a 768.007 billion yen net loss attributable to shareholders. The loss stems mainly from fund asset impairments and default losses on a US AI data center project, with free cash flow deteriorating to -5.9 trillion yen.
China Mobile announced at the 2026 Mobile Network Conference that its 2027 procurement will exceed 500 billion yuan, including over 100 billion yuan in compute investment, which it calls its largest growth area. The carrier is building three gigawatt-scale compute parks this year and two more next year for a total of seven, with smart compute capacity set to double for two consecutive years.
After the US midterm elections, investors worried about slower AI data-center expansion and dumped tech and AI hardware stocks. The Philadelphia Semiconductor Index fell about 12% in a single day, its second-largest one-day drop on record, while the Nasdaq slid 5.5% into correction territory and five tech billionaires including Musk lost a combined $152.9 billion.
Amazon, Microsoft, Alphabet and Meta disclosed at least $720 billion in combined capex for this year, mostly aimed at AI infrastructure, on top of at least $1 trillion in multi-year spending commitments. Yet the buildout remains no closer to delivering promised returns, and the plans have already triggered a sharp tech selloff on Wall Street.
After Ukrainian drone strikes damaged facilities housing AI supercomputers and disrupted cloud services, Russian tech firms are rushing to harden and bury data centers underground, while the Kremlin mandates reserved compute capacity for emergency state needs and even permits use of foreign data centers. The conflict is now directly reshaping the physical security and geography of compute infrastructure.
OpenAI, SoftBank Group, and SB Energy have partnered to develop multi-gigawatt AI data center campuses. The initiative includes a 1.2 GW facility in Texas designed to support the Stargate project's compute demands.
OpenAI and NVIDIA have announced a strategic partnership to deploy AI datacenters with a total power capacity of 10 gigawatts (GW), powered by NVIDIA systems. The first phase of this massive infrastructure project is scheduled to launch in 2026, signaling a deepening integration between the leading AI model developer and the dominant chip provider.