Reading Time : 3minutesMicrosoft says Copilot costs less than Claude and ChatGPT, citing a study it commissioned that puts Copilot at $30 a month for everyday users, compared with $81 for ChatGPT and $205 for Claude. The claim is plausible, but it rests on modeled workloads, list prices, and Microsoft-controlled assumptions. Anthropic’s own customer figures suggest the real-world gap may be much smaller.
Reading Time : 3minutesIsomorphic Labs has raised $5 billion to accelerate its effort to turn AI-driven protein science into real medicines. Backed by Sequoia, Greenoaks, and Alphabet, the company says its technology is approaching a pivotal test: moving its first drug candidates into clinical trials by the end of 2026.
Reading Time : 3minutesDeepSeek has quietly restricted its free chatbot, cutting off dormant long conversations, limiting message length, and imposing a daily usage quota. The company has not explained the changes, but the move follows repeated outages and points to a deeper issue: China’s constrained access to advanced AI computing power.
Reading Time : 3minutesTSMC received a record NT$167.6 billion in government subsidies in the first nine months of 2026, with almost all of it arriving in the third quarter. As its Arizona operation turned sharply profitable, the figures raise a larger question: how much of the chipmaker’s global expansion is being financed by taxpayers, and what do governments receive in return?
Reading Time : 3minutesOpenAI’s dismantling of its stand-alone robotics team signals a major strategic shift: the company now believes general-purpose frontier models, especially GPT-6 Astra, can control robots directly without specialized robotics AI. The reorganization moves researchers into core research and Applied teams, even as experts warn that real-world robotics still demands dedicated systems, deployment data, and reliable onboard intelligence.
Reading Time : 3minutesDario Amodei’s claim that AI could make cancer, Alzheimer’s, heart disease and mental illness “relics of the past” within a decade has drawn skepticism from drugmakers, physicians and biotech leaders. The dispute is not whether AI can help medicine, but whether a sweeping promise can outrun the slow work of proving a treatment safe and effective.
Reading Time : 3minutesAI agents were promised as the next way to order food, book rides and plan vacations. Yet Uber, DoorDash and Airbnb report almost no agent-driven activity, while eBay appears to be restricting access. The early numbers reveal a widening gap between the hype around autonomous shopping and what consumers actually want to do.
Reading Time : 3minutesSix governments and ten AI companies, including Anthropic, OpenAI and Microsoft, have signed the Paris Call for AI Safety and Security. The agreement commits them to shared frontier-model evaluations, faster incident warnings and clearer accountability. It arrives amid mounting cyberattacks and growing concern over powerful AI systems that may operate beyond effective human control.
Reading Time : 3minutesTaiwan’s market plunge reflects a global chain reaction: doubts about U.S. AI data-center growth hit chipmakers first, then Taiwan’s semiconductor supply chain, and finally the broader TAIEX. With TSMC dominating the index, a shift in AI expectations can quickly become a market-wide event, exposing both the strength and concentration risk of Taiwan’s technology-led economy.
Reading Time : 3minutesMistral says it needs far more compute to rival leading AI labs, exposing a widening gap between Europe’s AI ambitions and the chip, power, and data center scale of its American and Chinese competitors. Its plan depends on sovereign infrastructure, major financing, and a race against export controls.
Reading Time : 2minutesTesla’s Optimus production has begun in Fremont, but the company calls it an early ramp, not a finished commercial rollout. At the same time, YYForce plans to deploy its first humanoid robot at a Singapore hotel to test workflows, staff collaboration and operational value. The developments suggest hospitality may become an early proving ground for physical AI, though neither company has announced a Tesla partnership.
Reading Time : 3minutesAlphabet raised $12.4 billion from share sales and $28.7 billion from notes in the third quarter while buying back no stock. Capital spending reached $51.4 billion, exceeding operating cash flow and leaving free cash flow at negative $9.1 billion. The company is funding an aggressive AI infrastructure expansion as investors weigh whether future cloud and AI returns will justify the enormous upfront cost.
Reading Time : 3minutesAnthropic has filed to go public, positioning itself for what could become the largest IPO ever. The filing reveals explosive revenue growth, enormous losses, and hundreds of billions in compute commitments. As investors weigh a valuation approaching $2 trillion, the offering will test whether Wall Street believes the AI boom can justify its extraordinary price.
Reading Time : 3minutesJensen Huang argues that Nvidia GPUs could remain valuable for eight years or more, turning AI compute into collateral for data center financing. But as AI factories scale worldwide, electricity, grid access, cooling, and construction may matter even more than chip performance.
Reading Time : 3minutesTesla now says the first Cybercabs could reach customers by the end of 2026, but the company’s own executives suggest most will remain in Tesla’s hands. Meanwhile, unsupervised Full Self-Driving for private owners has slipped to around mid-2027, and only within limited Robotaxi zones. The promises are narrowing, even as the company talks about scale.