Sanjay Sahay

Two months ago, OpenAI’s leader Fidji Simo warned staff it was “code red” time because Anthropic was surging in enterprise (big business) AI use. Now, fresh spending data from fintech company Ramp flips the script — Anthropic has taken the top spot from OpenAI for the first time among paid business users. Ramp’s latest AI […]

Google kicked things off early before its big I/O event next week, unveiling major AI upgrades at the Android Show. They introduced AI-native Googlebooks laptops, a new Gemini Intelligence system across devices, a smart Gemini-infused mouse cursor, and more tools to unify AI in their ecosystem. The star is the new line of Googlebooks, shipping

Mira Murati’s Thinking Machines Lab has finally stepped into view with a clear message: the future of AI may not be only about autonomous agents running tasks alone, but about systems that collaborate with humans in real time. Its new “interaction models” are designed for voice, video, and text working together in a live, streaming

Google DeepMind has launched the “AI co-mathematician,” a smart team of AI agents that acts like a research partner for mathematicians. It handles everything from brainstorming ideas and hunting for old papers to running calculations and testing proofs. No more solo struggles—this tool creates a shared workspace where AI and humans collaborate step by step.

Anthropic needs more GPUs fast to power Claude, as demand outstrips supply and hits usage caps. Just months ago, Elon Musk called Anthropic “Misanthropic” and accused it of hating Western Civilization, yet now they partner up. SpaceXAI’s Colossus 1 in Memphis delivers full access to 300+ MW and 220K+ Nvidia GPUs within weeks, doubling Claude

AI’s massive hunger for high-bandwidth memory (HBM) and DRAM is vacuuming up global chip supply, sidelining consumer devices. Data centers from hyperscalers like Nvidia and Google prioritize high-margin AI servers, leaving factories churning out pricier laptops and vanishing budget PCs. Gartner predicts PC prices jumping 17% and smartphones 13% in 2026, with sub-$500 entry-level models

OpenAI has launched three new voice AI models: GPT-Realtime-2, GPT-Realtime-Translate, and GPT-Realtime-Whisper. These tools make AI better at handling real-time speech, reasoning, and tasks during live conversations. GPT-Realtime-2 adds strong reasoning like GPT-5, supports multiple tools at once, speaks while thinking, and controls tone for more natural talk. On the Big Bench Audio test, it

OpenAI is accelerating development of its first AI agent phone, targeting mass production in the first half of 2027—a year ahead of earlier estimates—driven by analyst Ming-Chi Kuo’s supply chain insights. The device features a specialized image signal processor with enhanced HDR for superior AI visual sensing, powered exclusively by MediaTek’s dual AI processors handling

AI needs massive data centers to grow, but building them on land sparks public anger over noise, land use, and power drain. Cities are pushing back hard. Peter Thiel’s startup Panthalassa offers a fresh fix: floating data centers in the open ocean. These 85-meter steel platforms bob on waves, turning ocean motion into clean electricity.

A Harvard study published in Science showed OpenAI’s o1-preview model outperforming two attending emergency room physicians across 76 real patient cases, achieving 67.1% diagnostic accuracy at triage compared to 55.3% and 50.0% for the doctors. Blinded reviewers couldn’t distinguish AI from human diagnoses, and the model even flagged a rare flesh-eating infection hours earlier than

The Google Enterprise Agent Platform marks a leap from “assisting tools” to full‑fledged agentic teammates inside every organization. Instead of humans doing the same repetitive tasks every day, specialized AI agents now observe, plan, and act on their own, updating systems while people focus on strategy, creativity, and relationships. This is not incremental automation; it