NAll news

All news

Saturday9 items

  1. HuxiuOpinion
    AI score66

    $1.65 Trillion in Off-Balance-Sheet Commitments and a 166-Fold Token Price Gap: Three Endgame Scenarios for the AI Capital Game

    The author reviewed the financial reports of the five major U.S. tech giants and found that more than 85% of Alphabet's and Amazon's quarterly profits come from unrealized gains on equity investments; their combined off-balance-sheet long-term purchase and lease commitments total approximately $1.65 trillion, already exceeding on-balance-sheet liabilities. Capital expenditure by the five giants is projected at $769.2 billion in 2026, while annual AI application revenue is about $200 billion, leaving a gap of roughly 5 times. After comparing China-U.S. pricing, the author notes that OpenAI charges $30 per million output Tokens while DeepSeek charges only $0.18, a price gap of 166 times, and concludes that the most likely endgame is that one node runs into trouble first and the problem then propagates through the chain of cross-shareholdings.

    #English text is being prepared.#Industry news#OpenAI

Friday12 items

  3. The PaperOpinion
    AI score58

    Behind hundreds of robot training grounds, embodied intelligence begins to recalculate the accounts

    The Paper reports that as of the end of June 2026, China had put more than 70 embodied intelligence training grounds into operation, with another 46 in the planning or construction stage. The first humanoid robot data training center in Beijing's Shougang Park had been running for less than 18 months when the original operator withdrew its own equipment and robots; the project side said the original teleoperation approach deviated from the new direction, and that data quality and precision fell short of expectations. The article notes that data collection centers are shifting from competing on production capacity to focusing on data sales rates, repurchase rates, and the capability improvements that data brings to models.

    #English text is being prepared.#Embodied AI#English text is being prepared.

  8. HuxiuOpinion
    AI score94

    An unreleased OpenAI model produced 722 mathematics manuscripts overnight, prompting a joint criticism from 25 Fields Medalists

    OpenAI published the openai/math repository on GitHub, disclosing 722 mathematics manuscripts and 372 result groups in one go, all produced by an undisclosed internal model, covering results such as a quasi-Riemann hypothesis, reducing the exponent of matrix multiplication over the complex field to 2.25, and a four-dimensional version of the Kakeya conjecture. The authors state that each result cost on average about 3 hours of ChatGPT Pro-level compute, whereas the September attack on the Navier–Stokes equations deployed roughly 10,000 agents, ran for 88 hours, and cost millions of dollars. 25 Fields Medalists signed a joint letter titled "The Serious Misalignment of AI in Mathematics," criticizing AI companies for treating problem-solving as a marketing tool, and the advisory group of the Institute for Advanced Study in Princeton demanded that such practices stop.

    #English text is being prepared.#OpenAI#Papers/Research

  9. HuxiuOpinion
    AI score58

    Data centers cluster in Ulanqab: Direct supply from the West Inner Mongolia power grid and Inner Mongolia's power surplus

    The delivered electricity price for data centers in Ulanqab is about 0.358 yuan/kWh, 0.442 yuan lower than Beijing. A single AI accelerator card saves about 2,700 yuan per year, and a 10,000-card cluster saves about 27 million yuan per year. The article argues that the low electricity price comes from the West Inner Mongolia power grid, as a provincial-level independent grid, allowing dedicated-line direct supply between data centers and power plants and exempting transmission and distribution tariffs, rather than from wind and solar resources themselves. Combined with Inner Mongolia's surplus capacity of more than 100 GW of wind power and about 130 GW of coal power as of the end of 2025, the region has already built or planned 89 data centers, with confirmed computing capacity of 15 GW. This institutional cost advantage is difficult for other computing hubs to replicate.

    #English text is being prepared.#English text is being prepared.#English text is being prepared.

Thursday5 items