News Square
Global Crypto News · Real-time Aggregation
📚 Deep Analysis
More →Bermuda taps WalletConnect to advance its onchain economy plan
Bermuda taps WalletConnect to advance its onchain economy plan
Former Swift CEO joins stablecoin firm OpenTrade as chairman
Former Swift CEO joins stablecoin firm OpenTrade as chairman
Mystery Wall Street Journal ad sparks new fight over tokenized stocks
Mystery Wall Street Journal ad sparks new fight over tokenized stocks
Crypto isn’t a partisan fight, proven by data
Crypto isn’t a partisan fight, proven by data
Fed leaves interest rates unchanged, as expected
Fed leaves interest rates unchanged, as expected
Coinbase earnings: Wall Street divided on trading rebound after Robinhood crypto revenue sinks 55%
Coinbase earnings: Wall Street divided on trading rebound after Robinhood crypto revenue sinks 55%
Crypto Long & Short: What happens when an AI agent pays for the wrong thing
Crypto Long & Short: What happens when an AI agent pays for the wrong thing
OpenAI Plans to Buy Google's AI Chips as It Looks Beyond Nvidia: Report
OpenAI plans to buy AI chips from Google, according to a person familiar with the matter who spoke with <i><a href="https://www.theinformation.com/articles/openai-longtime-nvidia-customer-plans-buy-googles-ai-chips" target="_blank" class="sc-adb616fe-0 dJryYS" rel="nofollow">The Information</a></i>. The step would loosen its dependence on Nvidia, the company that supplies most of the hardware behind the AI boom. Neither company has confirmed the plan, and the price, the number of chips, and the delivery dates have not been disclosed. OpenAI is one of Nvidia’s biggest customers. The chips OpenAI is reportedly after are Google's tensor processing units, or TPUs, processors designed for one job: the math that AI models run on. Google <a href="https://cloud.google.com/blog/products/ai-machine-learning/google-supercharges-machine-learning-tasks-with-custom-chip" target="_blank" class="sc-adb616fe-0 dJryYS" rel="nofollow">revealed the chip</a> in 2016, saying it had already run TPUs in its data centers for more than a year. Reports in June 2025 said OpenAI had begun renting Google's TPUs, and the company described it as early testing with no plans to deploy them at scale. A purchase would mean owning the hardware instead of paying Google by the hour. Other details of the plan have not been made public. A<a href="https://openai.com/careers/software-engineer-tpu-compilers-san-francisco/" target="_blank" rel="nofollow external noopener" class="sc-adb616fe-0 dJryYS">> job listing</a> from OpenAI in San Francisco asks for an engineer to lead the optimization of its frontier models on TPUs so the rumors have weight. The work involves compilers, the software that translates an AI model into instructions a specific chip can run, which is why switching chips is harder than swapping parts. Richard Ho, who leads OpenAI's hardware program, spent nearly nine years at Google and was one of the leaders of its Cloud TPU project before joining OpenAI. In September 2025, Nvidia signed a letter of intent to invest <a href="https://decrypt.co/340794/nvidia-100-billion-openai-historic-ai-infrastructure-deal-stock-pops" target="_blank" class="sc-adb616fe-0 dJryYS" rel="nofollow">up to $100 billion</a> in OpenAI as it deployed at least 10 gigawatts of Nvidia systems. A gigawatt is a billion watts, the unit data center builders now use to size their projects. Two weeks later, AMD gave OpenAI the right to buy up to <a href="https://decrypt.co/343068/amd-stock-skyrockets-openai-bet-loosen-nvidia-grip-ai-chip-market" target="_blank" class="sc-adb616fe-0 dJryYS" rel="nofollow">160 million AMD shares</a>, roughly 10% of the company. The right is tied to OpenAI deploying 6 gigawatts of AMD chips. Nvidia ended up putting <a href="https://decrypt.co/360195/nvidia-probably-done-investing-openai-anthropic-ceo-why" target="_blank" class="sc-adb616fe-0 dJryYS" rel="nofollow">$30 billion</a> into OpenAI's funding round, well below the $100 billion in the letter of intent. CEO Jens
Alphabet's AI search monetization problem may continue to weigh on its stock price.
According to analysts Justin Patterson and Sergio Segura, the commercialization progress of Alphabet's AI-powered search engine has fallen short of expectations and may continue to put pressure on the stock price in the short term. Revenue from Alphabet's Google Services business, including search, came in below market expectations in the third quarter, and the analysts said the results confirmed the concerns they had expressed before the earnings release. They noted that the disappointing search revenue performance could continue to weigh on the stock over the next one to two quarters, with investors waiting for signs to determine whether the issue is temporary or structural. Alphabet shares fell more than 10% in premarket trading.
Ant UBTree improves speculative decoding: parallel screening of draft trees lets large models verify more tokens at once.
According to news from Bijie.com, Ant Ling has open-sourced UBTree, a speculative decoding solution for large models. The solution works by filtering draft trees in parallel, allowing a small model to "draft" ahead of time, after which the large model verifies multiple tokens at once, thereby reducing the number of computation rounds required for token-by-token generation. UBTree improves the filtering method by first predicting in parallel the tokens that may appear at several subsequent positions, and then using a lightweight model to determine whether adjacent tokens can connect reasonably. After adjustments, the average speedup ratio for tree construction using the dflash 2 selector was 5.95x. After switching to the ubtree selector and adjusting the training method, it increased to 6.88x, a further improvement of about 16%.