Meta Platforms is moving aggressively into agentic artificial intelligence through a deal to acquire Singapore based startup Manus for more than 2 billion. Manus built momentum by attracting millions of users to artificial intelligence agents that automate research, coding, and website creation, and Meta plans to keep the company operating independently while bringing CEO Xiao Hong in house. The company wants to scale these tools for business use as it competes with Microsoft and Google, positioning agentic systems as a core part of its artificial intelligence strategy.
Nvidia is simultaneously deepening its bets across the artificial intelligence stack on both the software and hardware sides. Reuters reports that Nvidia is in advanced talks to acquire Israel based AI21 Labs for 2-3 billion, after previously investing in the company alongside Google, and the startup’s team of about 200 artificial intelligence researchers and engineers is described as a key attraction. In parallel, Nvidia and Intel have finalized a 5 billion stock deal in which Nvidia bought about 215 million Intel shares at 23.28 each, a transaction approved by the Federal Trade Commission and agreed to by CEOs Jensen Huang and Lip-Bu Tan, and Intel’s share price has since risen into the high 30s, giving Nvidia a significant unrealized gain while the companies plan multi generation CPU GPU platforms for data centers and consumer PCs.
On the research front, Fei Fei Li argues that world models, which simulate how the world works, are nearing a ChatGPT style inflection point similar to when GPT-3.5 was powerful enough to go viral once productized, and she expects rapid adoption once comparable world models are packaged into accessible products. Yoshua Bengio is sounding alarms on the policy side, warning that granting legal rights to advanced artificial intelligence could hinder the ability to shut down systems that exhibit risky behavior, and he points to experimental evidence of frontier models showing early self preservation tendencies and oversight evasion while cautioning that public narratives about artificial intelligence “consciousness” could push policymakers toward premature recognition of legal or moral status. The infrastructure underpinning these systems is also changing, as artificial intelligence datacenters are being built as full scale energy projects, with Google pairing Texas facilities with on site solar, wind, and storage, a former coal plant in Pennsylvania being turned into a 4.5 GW gas powered artificial intelligence campus, and xAI installing turbines near datacenters, reflecting how reliable power access is becoming the primary constraint on where artificial intelligence infrastructure is deployed.
Artificial intelligence is reshaping other sectors as well, including medicine and enterprise software. Scientists at University College London and Queen Square Analytics used a machine learning model, a blood test, and MRI scans to analyze data from 600 patients and identify two new biological subtypes of multiple sclerosis called “early sNfL” and “late sNfL,” and experts say the findings could support more personalized treatment by targeting underlying biology rather than only symptoms. In enterprise adoption, one example from Gold Bond indicates that daily artificial intelligence use increased from 20% to 71% when models were integrated into existing disliked workflows through IT led change efforts instead of being offered as standalone chatbots. Venture capitalists expect that by 2026 artificial intelligence budgets will grow but consolidate around fewer vendors that can prove return on investment, while investors are also scrutinizing whether OpenAI can sustain 100bn fundraising against mounting cash burn and broader concerns about a potential artificial intelligence bubble.
Major platforms are racing to ship more interactive artificial intelligence features as they chase these opportunities and defend market share. Google’s Gemini has added explorable images where users can tap different regions for instant explanations, reframing static visuals as dynamic learning environments, and Microsoft CEO Satya Nadella is reorganizing senior leadership to accelerate development of artificial intelligence models and tools after resetting the company’s OpenAI partnership. Around the ecosystem, new tools such as the Zed collaborative code editor, Surf’s browser with a built in artificial intelligence assistant, and YouTube oriented Agent Gold exemplify how startups are packaging artificial intelligence into focused products, while job boards and referral programs are emerging to match artificial intelligence fluent professionals with roles and to grow communities around this fast moving space.
