21 stories tagged ai infrastructure.
Langflow RCE flaws face active exploitation with over 15,000 successful attacks. Attackers harvest AI provider keys and cloud credentials from exposed instances.
Attackers are exploiting CVE-2025-62593 in Ray through Firefox and Safari. CISA gave agencies 72 hours to patch or stop using it.
Nvidia and six Wall Street firms aim to mobilize $500B for AI factory financing. GPU compute becomes an investable asset class, with 25% loan backstops.
Amazon's 7.65-gigawatt off-grid gas plant in Texas, permitted to emit 33 million tons of CO2 per year for AI data center power, is twice the output of the worst US coal plant.
SpaceX's AI revenue hit $2.6 billion in Q2 2026, nearly triple its space revenue. The rocket maker is now a neocloud selling compute to Anthropic and Google.
Enterprise GPU utilization sits at roughly 5%, six times worse than a no-effort baseline. GPU utilization is now the binding constraint for AI infrastructure.
AMD will invest up to $5 billion in Anthropic for up to 2 GW of MI450 capacity. Chipmakers are now buying equity in their own customers.
Memory costs are rising, but Apple's $29.6B profit makes the AI price hike harder to defend. The pass-through is a hardware margin test.
Stacking transistors, IBM's 0.7 nm nanostack research chip nearly doubles its 2021 density claim to almost 100 billion transistors.
ASML's $400M High-NA EUV tool brings an 8 nm lithography step for denser AI chips. Treat it as a roadmap risk and cost signal.
45C liquid cooling lets Rubin racks reject heat with dry coolers, cutting on-site water use while shifting scrutiny to power and buildout.
RAM price shock forced Nothing to skip a 2026 CMF phone, showing AI memory demand now decides what budget hardware can ship.
A 60-day FERC test for large-load interconnection means AI builders should treat power flexibility as product infrastructure.
MLPerf Training 6.0 shows Blackwell leading all seven tests, but the useful signal is scale: 8,192 GPUs and MoE training pressure.
Flexible data centers let AI sites cut load for a few peak hours, with Duke finding 76 GW of headroom at 0.25% curtailment.
The first AgentPerf results for concurrent AI agents put NVIDIA GB300 NVL72 at up to 20 times Hopper's efficiency per watt.
Amazon data centers used 2.5 billion gallons of water in 2025. Treat that disclosure as a roadmap risk for AI products.
Confidential inference lets Apple run AFM 3 Cloud Pro on NVIDIA GPUs in Google Cloud while keeping PCC privacy promises.
A 365-day pause on large new facilities: Seattle's data center moratorium makes 369 MW the warning label for AI roadmaps.
Shelbyville's $2B Prologis campus shows the AI data center backlash is now a zoning risk: trust can bottleneck compute before power does.
Google's 100 MW Voltus deal tests whether homes will flex for AI, using virtual power plants that coordinate flexible loads as capacity.