On October 21, 2026, Rohan Paul stated that Gemini 4 Argon processed approximately 25 trillion tokens in the first 72 hours after launch. He cited Logan Kilpatrick's statement and, based on that, calculated an average of about 96 million tokens per second. Rohan Paul also said that demand for intelligence is still underestimated. Currently, this claim appears only in a single-source report by Rohan Paul, with no other independent sources or official evidence to cross-verify it.
On October 21, 2026, Databricks announced the general availability of Databricks AI Runtime (AIR). According to information published on the Databricks Blog the same day, developers can train and fine-tune open models on serverless NVIDIA GPUs without having to manage GPU infrastructure themselves. AIR supports NVIDIA NeMo, NGC containers, and Nemotron cookbooks, as well as multi-node distributed training, with the underlying clusters and networking managed by Databricks. Trained models can be deployed for real-time inference through Databricks Model Serving, and access to training data is governed by Unity Catalog. This news currently comes from a single source, the official Databricks blog, and there are no other independent reports or subsequent developments yet.
On October 21, 2026, Databricks announced the general availability of Databricks AI Runtime (AIR). According to information published on the Databricks Blog the same day, developers can train and fine-tune open models on serverless NVIDIA GPUs without having to manage GPU infrastructure themselves. AIR supports NVIDIA NeMo, NGC containers, and Nemotron cookbooks, as well as multi-node distributed training, with the underlying clusters and networking managed by Databricks. Trained models can be d
On October 21, 2026, WorkBuddy published content reviewing its earlier process of using WorkBuddy to break down SpaceX's second-quarter financial report. The related post stated that the tool can complete a full professional financial report analysis workflow—including data extraction, segment breakdown, and valuation verification—within minutes. The post also mentioned that SpaceX achieved quarterly profitability for the first time in the third quarter, with revenue of $14.6 billion and AI revenue growth of 830%, and stated that the judgments about the direction of AI and Starlink made when using WorkBuddy to break down the Q2 data a week earlier are now being validated by Q3. At present, this review consists only of statements from the WorkBuddy publisher's side, and no other independent sources have cross-verified it.
On October 21, 2026, WorkBuddy published content reviewing its earlier process of using WorkBuddy to break down SpaceX's second-quarter financial report. The related post stated that the tool can complete a full professional financial report analysis workflow—including data extraction, segment breakdown, and valuation verification—within minutes. The post also mentioned that SpaceX achieved quarterly profitability for the first time in the third quarter, with revenue of $14.6 billion and AI reve
On October 21, 2026, Rohan Paul relayed survey data on large enterprises in the United States and the United Kingdom, covering software budgets and vendor lock-in. The survey showed that 28% of these enterprises' software budgets were spent on upgrades they would not have chosen themselves; 78% of respondents said that after software installation, the primary vendor decides when and how changes are made; and 95% had paid for upgrades they expected would make outcomes worse. The most recent vendor-driven change cost an average of $12.87 million per year. 34% of respondents cited vendor-driven costs as a major factor hindering AI or data projects, and 56% cited them as a moderate factor. Currently, this relay has only Rohan Paul as its source, with no other independent sources cross-verifying it and no subsequent developments.
On October 21, 2026, Rohan Paul relayed survey data on large enterprises in the United States and the United Kingdom, covering software budgets and vendor lock-in. The survey showed that 28% of these enterprises' software budgets were spent on upgrades they would not have chosen themselves; 78% of respondents said that after software installation, the primary vendor decides when and how changes are made; and 95% had paid for upgrades they expected would make outcomes worse. The most recent vendo
Rohan Paul offered the assessment that Meta's AI agent Muse reached 10.2 million downloads in the US within 40 days of launch. The data he cited also shows that historically only 11 apps have reached this pace, most of them mobile games. On retention, Muse's US 7-day retention rate is 30% and its 14-day retention rate is 23%, compared with competitor averages of 19% and 16%, respectively. On October 16, its daily active users reached 2.5 million, up 15% in a week, but still less than a quarter of its US downloads. At present, this assessment is seen only from Rohan Paul as a single source, with no other independent sources or official evidence to cross-verify it.
Rohan Paul offered the assessment that Meta's AI agent Muse reached 10.2 million downloads in the US within 40 days of launch. The data he cited also shows that historically only 11 apps have reached this pace, most of them mobile games. On retention, Muse's US 7-day retention rate is 30% and its 14-day retention rate is 23%, compared with competitor averages of 19% and 16%, respectively. On October 16, its daily active users reached 2.5 million, up 15% in a week, but still less than a quarter o
On October 21, 2026, Ethan Mollick announced that his new book "Co-Existence" was published that day. According to his introduction, the book is a guide on how to live and work in a world where "AI is sometimes better than you, but not always." Mollick also said that to accompany the book's release, he worked with Opus 5.5 to create a rather interesting experience for the book's official website. Currently, the verifiable information about the book's publication comes from this single representative source; there are no other independent sources or subsequent factual developments.
On October 21, 2026, Ethan Mollick announced that his new book "Co-Existence" was published that day. According to his introduction, the book is a guide on how to live and work in a world where "AI is sometimes better than you, but not always." Mollick also said that to accompany the book's release, he worked with Opus 5.5 to create a rather interesting experience for the book's official website. Currently, the verifiable information about the book's publication comes from this single representa
On October 20, 2026, the China Finance 40 Forum (CF40) released its CF40 annual research report titled "China's New Economic Paradigm" at the 2026 Bund Summit. The report notes that new economy increments have surpassed the traditional economy represented by real estate, but AI's productivity dividend has yet to be significantly realized. The report shows that in the first half of 2026, the central growth rate of AI-related import and export value year-on-year was about 60%, contributing more than half of import and export growth; Epoch AI estimates that Chinese models have lagged behind US frontier models by an average of about 7 months since 2023. Bai Chongen believes AI application is still at an early stage, and the volume invested at the application level is far from sufficient; he is optimistic about demand for data centers, but less optimistic about the profitability of model companies. Currently this report is covered by 36Kr AI Report, and there are no other subsequent factual developments.
On October 20, 2026, the China Finance 40 Forum (CF40) released its CF40 annual research report titled "China's New Economic Paradigm" at the 2026 Bund Summit. The report notes that new economy increments have surpassed the traditional economy represented by real estate, but AI's productivity dividend has yet to be significantly realized. The report shows that in the first half of 2026, the central growth rate of AI-related import and export value year-on-year was about 60%, contributing more th