Timo Mulder

Russia will bear the costs for the "Power of Siberia 2" gas pipeline itself and supply gas to China at domestic prices. These are the very terms the Chinese side presented to the Russian delegation during negotiations in May, according to the *Wall Street Journal*. Experts cited by the newspaper believe China intends to stall on the pipeline issue—effectively keeping Russia on the hook—until its position has deteriorated to the point where it is forced to accept any terms offered. "Putin’s war in Ukraine has not only dragged Russia into a quagmire but also handed Xi all the leverage needed to transform a relationship that was once a contest between near-equals into one of Chinese dominance across virtually every sphere," the *WSJ* concludes.

The M7 Ultra is designed to support as much as 1.5 terabytes of memory — double the capacity planned for the M5 Ultra — though whether Apple ultimately offers that configuration will depend on the state of the industry.

OpenAI burned $20.9 Billion in 2025 - and the tech industry's loudest skeptic just went on CNBC to call the entire AI boom "kayfabe" "This is a 10-to-30-Billion-dollar TAM industry pretending to be a trillion-dollar one" Ed Zitron - the critic who reported OpenAI's alternative financials alongside the FT - laid out the full bear case in 8 minutes: > The $20.9 Billion Burn (00:00): - Per the FT-reported financials, OpenAI's costs grow linearly with its revenue - "no amount of specialist silicon will bring these costs down" - and the IPO just slipped toward 2027 after a trillion-dollar valuation didn't materialize > The "Token Trap" Confession (01:04): - Palantir's Alex Karp, played back on air: enterprises "chillax and waste time with tokens... and they get my IP". Zitron goes further - LLMs can't charge on outcomes because they hallucinate, so the business model is built to "encourage waste" > The 7.1 Gigawatt Gamble (04:12): - Oracle is building 7.1 GW of capacity for ONE customer - and warned in its own annual report it might not get paid. Zitron estimates OpenAI would need ~$75B a year in compute spend for Stargate alone > The Rent-Back Red Flag (04:54): - Per The Information, Nvidia will pay neoclouds to rent back their own GPUs - "something that only happens in an industry without diverse and real demand" > The Missing Number (06:00): - Microsoft, Amazon, Google and Meta report AI "run rates" but never AI revenue "Public companies love good news. If they had good news, why wouldn't they share it?" > The First Domino (06:43): - A Goldman analyst says the first hyperscaler to cut CapEx "gets rewarded by the markets" - and in an industry of followers, everyone copies whoever blinks first Bookmark & Watch Now ↓