#AI Infrastructure

Anthropic is the customer in RUM Group's $13.7B GPU deal, The Information reports. The deal comes with a penny warrant for 50.8M RUM shares.

SoftBank shares fell 11.6% in Tokyo after OpenAI's Altman backed slowing AI and ruled out a 2026 IPO. SoftBank has committed $64.6 billion to OpenAI.

SK hynix fell 5.4% and Samsung 3.4% at Monday's open, dragging the KOSPI down 3.2%, as Asia reacted to Anthropic's AI slowdown call, oil and a Fed hike.

Anthropic told investors it expects a second straight quarter of adjusted operating profit, the FT reports. Its 80%+ gross margin excludes training costs.

Nasdaq 100 futures fell 1.2% at Sunday's open, twice the S&P 500, after Anthropic's call to slow AI. Oil rose above $103 and a Fed hike looms Wednesday.

Data center construction costs about $6M–$13M per MW across the USA, EU, UAE, India and Pakistan, and a 10 MW power bill runs $8.7M–$26.3M a year.

Microsoft revealed Azure's $101.9B revenue for the first time, and retired the segment whose margin fell from 64% to 57%. What Microsoft stock investors lose.

Nvidia now gives some customers up to a year to pay, guarantees $105B of an OpenAI campus's rent and holds $94B of stakes. Why a 106% grower trades at 24x.

SpaceX's Q2 filing shows $15.83B of AI capex against $1.17B on Space — 86% of all capital spending. Only Starlink earns. What SPCX investors are actually buying.

Amazon's Q2 net income rose 245%. A $50.5B unrealised markup on its Anthropic stake drove 62% of pre-tax profit — and it flatters the P/E. What the 10-Q shows.

We ranked 41 US-listed AI stocks by Q3 performance. Enterprise software is up 33.5% at the median, semiconductors down 14.6%, and no chipmaker made the top nine.

Oracle opened at $164.43 and closed at $150.28. The release explains why: record AI backlog, negative free cash flow, and $20 billion of stock sold in the quarter.

Reuters reports Nvidia is in talks to invest in Anthropic’s IPO, which could raise $100 billion at a $2 trillion valuation. The terms remain tentative.

Firebird opened its Armenia AI factory, but the 70,000-GPU, 300 MW vision still depends on power, financing and customers. Here is what is actually live today.

Microsoft's expanded multibillion-dollar Mistral pact reverses their 2024 roles — Microsoft now buys European GPU capacity and ships Mistral's models to regulated firms.

NAVER, NVIDIA and Brookfield are tripling Korea's AI factory to 200MW. The real story isn't the megawatts — it's the ~$10 billion capital stack behind sovereign AI.

Alphabet posted record Q2 2026 net income and its first negative free cash flow together. Why $44.9B of AI capex — not the profit — is the real story.

Upper90 will lend General Compute up to $400 million secured by SambaNova inference chips, not Nvidia GPUs. Why inference silicon as loan collateral is an untested bet.

Fireworks AI raised $1.505B at a $17.5B valuation. The number that matters is 95%: nearly all its tokens come from customer-specialized models, not raw open weights.

Google's Ironwood TPU has been generally available since April, yet there's still no public per-hour price. What that silence reveals about AI inference costs.

AMD and Cerebras will split AI inference across Helios racks and the Wafer-Scale Engine. The 5x per-watt claim is modelled against Cerebras-only hardware.

CoreWeave measured 10× more DeepSeek-R1 tokens per megawatt on Vera Rubin NVL72, but missing methodology limits what NVDA investors can conclude today.

Xinference 3.0 turns database-backed authentication on by default and adds an air-gapped deployment path. Here is what operators must test before upgrading.

GPT-5.6 is on Amazon Bedrock, but pricing, context limits, caching and contract terms determine whether AWS commitments create real enterprise savings.