Nvidia reported quarterly profit of $43 billion and trailing 12-month profit of $120 billion, quarterly revenue of $68.1 billion, and said AI data-center chip sales rose 71% to $61.7 billion while it controls about 90% of cutting-edge AI semiconductors. - The New York Times

Nvidia received a U.S. license to export a small number of H200 chips to China. - Bloomberg

Move fast and burn out: Hieu Pham, a former xAI and OpenAI engineer, said he quit due to burnout and is returning to Vietnam with his family to recover after joining xAI in August 2024 and moving to OpenAI in August 2025. - Business Insider

OpenAI launched Frontier earlier this month, an enterprise platform for role-based AI agents with early adopters including HP, Intuit, Oracle, State Farm and Uber, and formed a Frontier Alliance with Accenture, BCG, Capgemini and McKinsey to deploy and certify it. - The Register

OpenAI hired Ruoming Pang, who left Meta last week after about seven months and had led AI infrastructure for Meta Superintelligence Labs and previously led Apple's models team. - The Information

Ni-hao propaganda:OpenAI said ChatGPT refused requests to assist Chinese influence operations. - Bloomberg

Amazon plans to invest up to $50 billion in OpenAI, with $35 billion contingent on an IPO or an artificial general intelligence milestone, according to three people involved. - The Information

Lowe's partnered with OpenAI to roll out the Mylow AI assistant in March 2025, which Marvin Ellison said drove about a two percentage-point in-store satisfaction increase and doubled online conversion rates. - Business Insider

Regulate this, law nerds: Anthropic-funded super PAC Jobs and Democracy has spent roughly $270,000 and plans nearly $700,000 to support Rep. Valerie Foushee's NC-4 re-election, raising concerns about AI-industry influence on a House AI commission. - The American Prospect

Anthropic's Claude, its flagship large-language assistant, has rapidly gained adoption among policymakers, financial institutions and tech firms across Washington, Wall Street and Silicon Valley, increasing Anthropic's market influence and shaping regulatory and enterprise AI discussions. - Axios

Anthropic acquired Vercept Inc. for undisclosed terms, will sunset Vercept's Vy desktop app within 30 days and recommend switching to Claude, and said the purchase will help integrate Vercept's features into Claude Cowork to enhance computer-use capabilities. - SiliconANGLE

Check Point researchers Aviv Donenfeld and Oded Vanunu found three vulnerabilities in Anthropic's Claude Code that allowed remote code execution and API-key exfiltration via repository-controlled config files; Anthropic issued fixes and published advisories and CVEs. - The Register

Anthropic accused three Chinese firms of using 24,000 fraudulent accounts to generate about 16 million exchanges to distill its Claude model, and Google and OpenAI reported similar global extraction attempts that raise IP, security, and market-concentration concerns. - New York Magazine

Anthropic retired Claude Opus 3 on January 5, 2026 via a new retirement‑interview process, will publish Opus 3's weekly essays in a newsletter, and kept access for paid claude.ai subscribers or API request, citing linear cost and capacity limits. (?) - Claude

Jack Clark, Anthropic's head of policy, said he limits his toddler's access to YouTube's algorithm, warned AI will require stronger parental controls, and advocated age verification, family profiles and algorithmic limits to protect children. - Business Insider

Cloudflare engineering director Steve Faulkner used Anthropic's Claude via the OpenCode agent to implement 94% of the Next.js API in one week for ~$1,100, producing open-source Vinext (Vite-based) reporting up to 4.4× faster builds and ~56% smaller client bundles. - The Register

xAI's Grok signed a Pentagon deal to operate inside classified networks for intelligence analysis, weapons development and battlefield planning, replacing Anthropic's Claude after Anthropic declined the Pentagon's "all lawful use" requirement. - GlobalSecurity.org

xAI bought a dormant Southaven, Mississippi power plant, deployed 27 temporary methane turbines to power an AI data center amid a planned $20+ billion investment and application for 41 permanent turbines, prompting resident complaints, legal, regulatory and permit disputes. - NBC News

The FBI used a search warrant to obtain Grok chat logs from X in the Simon Tuck investigation, asserting the prompts generated approximately 200 pornographic videos of a woman resembling the victim's wife. - 404 Media

DeepSeek withheld pre-release access to its V4 model from U.S. chipmakers Nvidia and AMD while granting early access to Chinese firms including Huawei, and its models have been downloaded more than 75 million times on Hugging Face. - Reuters

GPUaranteed loans: Nvidia said it provided $3.5 billion in guarantees for companies leasing land, power and data center facilities—four times last quarter's amount—to help early-stage, unrated firms secure power contracts and debt financing for data-center buildouts. - The Information

Tech companies are increasingly using chip-backed loans—secured by semiconductor assets—to raise capital for AI investments, attracting investors with higher yields while shifting debt off balance sheets and expanding asset-backed financing for semiconductor supply chains. - Financial Times

Datacenter cancellations rose to 26 through January from one in October, while transformer lead times stretched to as long as four years and costs rose up to sixfold, prompting PJM to defer connections into 2030 and investor retreat. - Yahoo

Citadel Securities' 2026 analysis says AI deployment is constrained by rising marginal compute, data-center and energy costs, with AI capex about 2% of GDP (~$650bn), about 2,800 planned US data centers, and adoption following an S-curve. - Citadel Securities

Citrini Research's viral note projected a 2028 AI-driven obsolescence of many white-collar roles, prompting a late-February stock wobble as investors reassessed valuations across payments, delivery and software, while experts warned near-term disruption is overstated. - The Economist

Pantheon Macroeconomics says AI capex accounted for nearly a fifth of the 2.2% year‑over‑year U.S. GDP increase in Q4 2025, while $3.8 trillion in tech‑stock wealth boosted consumption and, with AI capex, drove about one‑third of growth. - Fortune

The Norway Government Pension Fund Global said artificial intelligence is detecting risks that other approaches are missing. - Bloomberg

Internal documents show Amazon's ProServe is shifting from hourly to outcome-based pricing—targeting up to 75% fixed-price projects—and reorganizing around 'agentic AI' as 'digital employees,' while aiming 20% yearly revenue-per-employee growth by 2026. - Business Insider

Drone Ranger Maver-AI-ck: Shield AI's Hivemind AI pilot flew Anduril's Fury (YFQ-44A) in a first flight test over the Mojave Desert, completing required mid-mission updates and maneuvers as part of the Air Force's Collaborative Combat Aircraft program. - Business Insider

Consultancies are winning mandates as Europe's defence spending post-Ukraine fuels rapid drone and capability scaling, exemplified by a 20‑drone AI swarm trial in Cumbria, while Source Global Research forecasts UK defence consulting revenues of £1.6bn in 2026. - Financial Times

Mexico's Jalisco New Generation Cartel (CJNG) has adopted AI, drones, and social media to enhance surveillance, logistics, decision-making, and recruitment, increasing cartel resilience and creating demand for counter-drone, cyberdefense, and intelligence tools. - WIRED

Alphabet-owned Intrinsic, spun out of X in 2021 and led by CEO Wendy Tan White, is joining Google as a distinct entity to work with Google DeepMind and leverage Gemini models and Google Cloud; financial terms were not disclosed. - TechCrunch

Skild AI CEO Deepak Pathak says his company built a general-purpose 'brain' controlling humanoid and dog-like robots that adapts to limb loss, raised $1.4 billion at over $14 billion valuation, and grew enterprise revenue to tens of millions in 2025. - Sources

Don’t worry by SaaS-y: Salesforce CEO Marc Benioff dismissed "SaaS-pocalypse" fears that AI will overtake business software while the company projected revenue below Wall Street expectations, prompting a drop in Salesforce shares. - Financial Times

Workday, shorter workweek: HR provider guided that subscription revenue growth will slow, citing mounting worries about artificial intelligence. - The Wall Street Journal

C3.ai said in a filing it will cut about 26% of its workforce; it had roughly 1,200 employees as of April 2025, and its shares fell over 20% after hours on weaker fiscal-year guidance. - Bloomberg

AMD will invest $150 million in Nutanix stock and commit up to $100 million for joint engineering and go‑to‑market to build a full‑stack AI infrastructure platform, while Nutanix reported Q2 revenue of $723M and ARR of $2.36B. - The Register

Economist Impact surveyed 639 senior decision-makers across five cities and found only 4% of businesses achieved ROI on AI, with 38% having an AI budget, 16% offering structured training and 8% having comprehensive AI governance. - The Register

Artificial intelligence complicates longstanding internet privacy risks by enabling broader data aggregation, reidentification, and automated profiling. - The New York Times

MATS Research with ETH Zurich and an Anthropic co-author showed LLM agents deanonymized 226 of 338 Hacker News users (67% recall) at 90% precision in a controlled experiment, costing about $2,000 (≈$1–$4 per profile). - The Register

An analysis of more than 1,000 YouTube Shorts recommended to children found the algorithm often promoted bizarre, frequently nonsensical AI-generated videos, and many lacked disclosure that they were AI-created. - The New York Times

Alvi Choudhury, 26, was arrested and held nearly 10 hours after Thames Valley Police's use of Home Office‑procured Cognitec facial‑matching software linked him to a burglary 100 miles away, prompting his lawsuit and police acknowledgment of possible algorithmic bias. - the Guardian

Tinder for AI snitches: Mary Inman of the nonprofit Psst said its encrypted "Tinder for Whistleblowers" matches anonymous deposits and unlocks keys only when multiple submissions align, offering legal and psychosocial support to global AI workers after recent safety-related resignations at major labs. - Rest of World

University of Melbourne/KPMG survey found 47% of professionals admitted inappropriate workplace AI use and 63% saw others doing so, while many reported unauthorized use, sensitive-IP uploads, reduced effort, and increased mistakes. - Quartz

IMF analysis estimates AI will affect about 40% of jobs worldwide, and workers across editing, translation, marketing, healthcare and mathematics report being asked to train or correct AI, leading to lower pay, reduced quality and some layoffs. - the Guardian

European Central Bank President Christine Lagarde said the ECB will be extremely attentive to job cuts driven by artificial intelligence. - Bloomberg

Some Chinese women are forming romantic relationships with AI chatbots as China faces a shrinking population and low birthrate, prompting Beijing to tighten AI regulations. - The New York Times

Perplexity launched Perplexity Computer, a general-purpose digital worker that orchestrates tools across 19 AI models and runs long workflows, available today on the web for Perplexity Max subscribers ($200/month) with Pro ($20/month) and Enterprise rollouts planned. - The Deep View

Feb. 24 Perplexity told SDNY court that Dow Jones and the New York Post 'cherry-picked' and 'badgered' its AI with hundreds of prompts in a 2024 copyright suit over alleged reproduction of paywalled reporting, a dispute that could set precedent. - Business Insider

Can an OpenClaw AI agent write compelling Fast Company stories? - Fast Company

AI teams use four to five disconnected agent tools—tracing, evaluation, guardrails, prompt optimization, simulation and monitoring—causing production failures in integration gaps and driving demand for consolidated, open, auditable full-lifecycle agent platforms. - Hacker Noon

At The Pragmatic Summit, 92% of developers use AI coding tools monthly; AI-native teams can achieve 2–5x throughput, some orgs report ~2x more customer incidents while healthy ones see ~50% fewer, and MIT's July 2025 study noted profitization drop-offs. - The Pragmatic Engineer

An AI chatbot promotion in China generated about 10 million orders in nine hours and roughly 120 million orders over six days, autonomously placing purchases and causing store shortages, delivery delays and server failures, revealing an AI-driven digital-physical throughput gap. - Hacker Noon

Adobe added a Quick Cut tool to Firefly that automatically generates a rough video cut from uploaded footage and a text prompt, aiming to speed first-draft editing for reviewers, podcasters and hobbyists while preserving editors' storytelling role. - Digital Trends

Follow the latest AI headlines via druce.ai on Bluesky