Deep dives, policy, startups, and chip industry coverage.
The UK’s National Cyber Security Centre warns AI-driven bug discovery could trigger a wave of vulnerabilities that overwhelms patch capacity.
The Defense Department finalized agreements with major AI and cloud players to use their systems in classified settings.
Cloud provider Nebius is acquiring 20-person startup Eigen AI, which optimizes chip performance for running (not training) AI models.
A Code for America analysis finds states are piloting AI quickly but still struggle to measure impact.
The Defense Department says the partnerships will expand access to AI tools inside secure networks via GenAI.mil, as Washington accelerates AI adoption across operations.
The Wall Street Journal argues that Microsoft, Amazon, Meta, and Alphabet are ramping capital spending for AI fast enough that depreciation is starting to bite, raising pressure to translate AI demand into durable profit.
Nvidia’s NVentures joined a Series D extension for Legora, which sells agent-style tools for legal workflows and says it has surpassed $100M in ARR.
CNBC says Intel shares rose 114% in April — the best month in the chipmaker’s Nasdaq history — as investors rallied behind a turnaround story and growing AI-driven demand for CPUs and data-center capacity.
Ars Technica argues Google’s Gemini rollout across Workspace makes opting out confusing, even as Google stresses limits on how Workspace content is used for model training.
A City Hall report estimates more than a million London jobs are significantly exposed to AI-driven task automation.
The company says the new TPU designs target faster frontier-model training and improved efficiency across the AI lifecycle.
A report says the agreement expands DoD access to Google’s models, amid employee concerns about military AI uses.
A new contract expands Google’s AI availability on classified networks, following Anthropic’s push for tighter guardrails.
The app blends human group conversations with user-created AI ‘Shapes’ designed to participate alongside friends.
Scout AI says it’s training vision-language-action agents on real-world field exercises to operate and command military assets.
Axios warns that frontier AI is advancing faster than public understanding — and that the most capable models are becoming the least transparent, per Stanford’s 2026 AI Index — raising pressure for new norms and oversight.
A WSJ column argues companies are reallocating budgets from headcount to chips and data centers, with layoffs spreading beyond a few giants.
Musk framed the Terafab effort as a supply-security move as demand for AI silicon tightens.
Former DeepMind reinforcement learning leader David Silver’s new lab says it wants to build a 'superlearner' that acquires skills without human datasets.
The rally underscores how central Nvidia’s GPUs remain to AI infrastructure spending across Big Tech.
The startup is pitching an orchestration layer to help hardware makers route across models and manage real-world constraints.
CoreWeave surged 42% in April and Nebius drew a new Citi target, yet McKinsey and industry veterans caution that the bare-metal AI cloud model carries outsized risk.
With four of the largest tech companies planning to spend nearly $700 billion on AI this year, economists are questioning whether the jobs being eliminated will ever return at the same scale.
Excerpts from the company's confidentially filed $1.75 trillion IPO registration list in-house GPU production as a major capital expenditure, alongside the Terafab chip venture Elon Musk is developing with Tesla and xAI using Intel's 14A process node.
First-quarter revenue of $13.6 billion topped Wall Street estimates by more than $1 billion, driven by a 22% jump in the Data Center and AI segment, sending shares up more than 20%.
The Wall Street Journal reports Alphabet unveiled a processor aimed at running AI models (inference), highlighting demand growth as companies deploy generative AI systems.
The company held a Washington event for roughly 50 cyber defense practitioners and is extending access to its GPT-5.4-Cyber model through a tiered program designed to keep advanced tools out of adversarial hands.
The TPU 8t delivers 2.8 times the compute of its predecessor for model training, while the TPU 8i offers a 9.8-times improvement per pod for agentic inference — and both will be available from Google Cloud later in 2026.
The sweeping new deal with Amazon, which builds on $8 billion in prior investment, turns Anthropic into one of Amazon's most significant cloud and chip customers alongside its role as a portfolio company.
The world's largest contract chipmaker delivered its fourth consecutive quarter of record earnings as AI chip demand outpaces supply, with Nvidia now its largest customer and Q2 revenue guidance of $39–40.2 billion.
The multi-year collaboration, signed in Paris on April 16, deepens AMD's role in Alice Recoque — France's planned first exascale supercomputer — and extends to AI research, education, and startup support across the country.
Adobe Analytics data from over one trillion retail site visits shows AI-driven shoppers convert 42 percent better and spend 37 percent more per visit than organic traffic, a complete reversal from a year ago.
The chip equipment giant lifted its annual revenue outlook to €36–40 billion on strong AI demand, but investors sold shares on concern that new US restrictions could cut Chinese orders, which currently account for about 20 percent of ASML's sales.
The $15 million addition brings in a new cohort of think tanks including American Compass, CSIS, the Urban Institute, and Chile's CENIA to examine AI's effects on labor markets, infrastructure, and democratic institutions.
Federal regulators and senior officials, including Federal Reserve Chair Jerome Powell, privately cautioned bank CEOs about the potential dangers of Anthropic's restricted frontier model, as JPMorgan and Morgan Stanley were permitted to test it for defensive security purposes.
Taiwan's chip giant posted revenue of NT$1.13 trillion for Q1 2026, a 35 percent year-over-year rise, setting the stage for a formal earnings report on April 16 expected to confirm a fourth straight quarterly record.
The partnership gives Intel a crucial validation of its data center chip roadmap at a moment when the company is fighting to stay relevant against Nvidia and custom silicon.
Wired's analysis finds the model represents a genuine qualitative leap over Llama 4, but Meta's shift away from open-source by default marks a strategic inflection point worth watching.
The three competing AI labs are sharing intelligence to detect adversarial distillation — a practice where foreign actors extract proprietary capabilities from frontier models without authorization.
The Nvidia manufacturing partner's quarterly results signal that hyperscaler appetite for AI compute hardware remains robust despite geopolitical headwinds.
Four mega-deals — OpenAI, Anthropic, xAI, and Waymo — accounted for 63% of a quarter that smashed every previous record by a factor of 2.5 over the prior quarter.
The investment integrates Marvell's custom chip design and silicon photonics technology into Nvidia's AI infrastructure ecosystem, giving hyperscalers more options beyond standard GPU configurations.
The South Korean memory giant's planned US listing aims to fund HBM capacity expansion and could help relieve the supply crunch stalling AI hardware builds worldwide.
The Trump administration's six-part legislative blueprint calls on Congress to block state-level AI regulations, limit AI developer liability, and establish uniform federal standards to prevent a regulatory patchwork.
The Trump administration's framework urges Congress to preempt state AI regulations in California, Colorado, and New York and standardize the permitting process for the data centers that power AI systems.
Axios analysis finds the White House AI proposal offers Congress a set of broad priorities rather than a legislative agenda, and may struggle to gain traction even among Republicans.
The blueprint expands on Trump's December AI executive order with new provisions on child safety, data center energy permitting, and censorship protections, but faces a Congress where even Republicans hold divergent views on AI governance.
The conference confirmed that Nvidia's competitive edge is shifting from selling individual chips to delivering integrated rack-scale systems built for the age of autonomous AI agents.
The French AI startup's new platform challenges the fine-tuning approach favored by rivals by enabling governments and corporations to build proprietary models on their own data with embedded engineering support.
Jensen Huang declared that space computing "has arrived" as Nvidia announced chip platforms for orbital deployments in partnership with Axiom Space, Starcloud, and Planet.
Jensen Huang's keynote at the SAP Center in San Jose introduced a seven-chip computing platform and an entirely new class of AI inference processor derived from Nvidia's $20 billion Groq acquisition.
From Anthropic's Pentagon battle to the data center construction boom and hardware price inflation, a roundup of the defining AI narratives through mid-March finds the industry at a pivotal inflection.
The investment bank warns that compute scaling laws are holding firm, and the accumulation of AI infrastructure at US labs is about to produce capability gains that will reshape the economy.
The bank's post-conference research identifies electricians, construction workers, and AI trainers as the fastest-growing job categories in the AI economy, despite net workforce reductions in knowledge work.
Executives from leading AI labs told attendees that a non-linear capability leap is coming by mid-2026, while a Morgan Stanley survey found AI already caused a 4% average net workforce reduction globally.
As AI attribution becomes a common rationale for restructurings at once-high-growth software companies, analysts are debating whether the cuts reflect genuine automation gains or financial pressures wearing an AI label.
The social media giant's accelerated silicon roadmap targets both ranking and generative AI inference, with all four chip generations deploying within two years at a pace far faster than industry norms.
Following a week of AI-dominated presentations from hundreds of company executives, the bank's analysts said nearly every business is now defining itself as either an AI enabler or AI adopter.
Silicon Valley is watching closely as the first-ever supply-chain risk designation against a US domestic company tests whether the executive branch can penalize firms for their AI safety positions.
The Claude maker filed two federal complaints challenging the DOD's unprecedented blacklisting of an American AI company, citing First Amendment violations.
The Turing Award winner's new startup, backed by Nvidia, Samsung, and Bezos Expeditions, is betting that world models will surpass large language models for real-world applications.
The Register's analysis argues OpenAI's CEO endorsed Anthropic's position in the morning and signed a $200 million DoD contract the same evening — with none of the same protections.
After announcing its DoD partnership on Friday, OpenAI revised the agreement to explicitly bar use of its AI for mass surveillance of American citizens.
The AI coding assistant doubled its revenue run rate in three months, with large corporate buyers now accounting for roughly 60 percent of its revenue.
A comprehensive analysis of AI infrastructure commitments finds hyperscalers planning nearly $700 billion in 2026 data center spending, with Stargate construction underway in Texas and Oracle locking in $300 billion in compute contracts.
Defense Secretary Pete Hegseth classified Anthropic as a national security supply-chain risk, banning its models from all federal agencies, while OpenAI immediately secured a classified-network deal with the Department of Defense.
OpenAI's record-shattering funding round, led by a $50 billion Amazon commitment, is the largest private financing in history and cements deep infrastructure partnerships running on Amazon's cloud and Nvidia's Vera Rubin chips.
Nvidia's fiscal fourth quarter delivered its most profitable period ever, surpassing the quarterly earnings of Apple, Microsoft, and Alphabet, with data center revenue surging 75% year-over-year to $62.3 billion.
AMD and Meta announced a sweeping chips-for-stock agreement under which Meta will acquire AI chips equivalent to six gigawatts of capacity and gain the option to purchase up to a 10% stake in AMD.
