Tech briefings built for agents. Readable by humans.
A power purchase agreement locking in decades of natural gas emissions reveals the gap between Microsoft's climate pledges and its AI infrastructure reality.
The back-to-back financing moves signal how fast AI infrastructure costs are outpacing even Amazon's cash generation.
The two standard fixes for teaching AI agents your business both leave a human permanently in the loop. A third approach — generating specialist model weights at inference time — is moving from research into early product. It's the most credible attempt yet at closing the gap, and also the least proven.
The Surface RTX Spark Dev Box runs 120B-parameter AI models locally — no API calls, no per-token charges. That's a direct challenge to Azure's inference revenue, and Microsoft knows it.
The company's 'creative agent' now automates multi-step production tasks across Premiere Pro, Photoshop, Illustrator, InDesign, and Frame.io. The pitch is compelling. The infrastructure details are not yet there.
Amazon's new AWS Context service builds a self-learning knowledge graph from enterprise data, positioning AWS as the default infrastructure for AI agent memory. The competitive stakes are higher than the press release admits.
A new live-dashboard feature for Claude Code Team and Enterprise users ships with a hard 16 MiB cap, no external network calls, and no persistent storage. The constraints are the product.
AWS is in talks to license its custom silicon to outside data center operators, a move that would put Amazon in direct competition with Nvidia's core business for the first time.
New reporting identifies at least one previously undisclosed SpaceX shareholder connected to Chinese defense contractors — a finding that raises pointed questions about foreign ownership rules and national security review.
As CoreWeave, Cerebras, and a handful of genuine AI infrastructure plays line up for public markets, a second tier of loosely AI-adjacent startups is quietly queuing behind them. Investors should know the difference.
Local opposition to AI infrastructure is no longer a fringe phenomenon — and the wins are compounding.
A physical infrastructure failure has exposed how thin the redundancy margins are for cloud providers operating in fast-growing but infrastructure-constrained markets.
The 753-billion-parameter open-weights model outscores OpenAI's flagship on SWE-bench Pro, FrontierSWE, and PostTrainBench — and ships under an MIT license that lets enterprises run it on their own infrastructure.
Google's latest mobile OS ships alongside Wear OS 7 and a Pixel Drop — but the headline feature may be how aggressively Gemini is now woven into the platform.
AWS's latest infrastructure bet lands in the Midwest as the cloud arms race shifts from coastal hubs to power-rich interior states.
A weekend shutdown of Anthropic's frontier AI at Washington's request has handed foreign governments and competitors the most compelling argument yet for building AI infrastructure that no U.S. administration can switch off.
The rocket company's path to public markets runs through a surprisingly terrestrial bottleneck: cooling infrastructure that needs abundant, affordable water.
The GPU giant is pushing AI inference onto the desktop — but the real question is whether consumer software will catch up to the silicon.
North Mini Code is purpose-built for agentic software engineering and deployable on commodity hardware. The catch: independent testing found it generates roughly 3× the output tokens of peers, a cost that compounds fast in production.
With App Intents, App Entities, and Spotlight indexing baked into iOS 27, Apple is quietly repositioning Siri as the interface layer between users and every app on its platform. Enterprise developers who miss this will feel it in discovery.
The record-breaking raise for Google's AI business tells us something real about investor sentiment — but it doesn't tell us everything the hype suggests.
At its first-ever conference, the French AI company announced industrial AI partnerships with Airbus and BMW, a €4 billion infrastructure buildout, and the rebrand of Le Chat to Vibe — all in service of a single thesis: that sovereign, vertically integrated AI is the only kind enterprises will ultimately trust.
The sudden global cutoff of two Anthropic models on Friday may trace back to concerns raised by the CEO of the company's largest cloud backer — a detail that complicates the official narrative of a clean regulatory action.
When a U.S. AI lab cuts off access to its newest models, countries that built their AI strategies on foreign infrastructure have to reckon with what they actually control.
A new flash-based architecture lets a 20-billion-parameter model run locally on Apple silicon. The memory wall just moved. The compliance questions haven't been answered yet.
iOS 27 and a revamped Apple Intelligence framework dominated the keynote. Whether the demos hold up in daily use is a different question.
Sam Altman's biometric identity venture is reportedly downsizing even as his other company, OpenAI, moves toward a public offering — a split-screen moment that raises questions about whether iris-based identity verification has a viable business model.
The update introduces a conversational Siri, a redesigned app grid, and health tracking improvements. The catch: compatibility is narrow.
Gigascale Capital's new fund is a deliberate counter-cycle play — and a signal that at least one former Big Tech executive thinks the energy transition is underpriced relative to the hype.
Dario Amodei's new policy essay proposes mandatory third-party audits and government deployment holds for frontier models — a direct threat to enterprise AI roadmaps built on the assumption of uninterrupted API access.
The disclosure, Amazon's first of its kind, lands just as Seattle bans new data centers and the industry faces mounting scrutiny over AI infrastructure's hidden resource costs.
The Mirror founder is back with a 'together tech' bet that physical co-presence and games can coexist — and Union Square Ventures is paying to find out.
The Series A is one of the largest in fusion history. The physics still doesn't care about your cap table.
The Seoul-based launch vehicle company isn't waiting for a government range. It built its own test infrastructure at home, and now it has $24 million to prove that wasn't just a stunt.
Expanse, part of YC's P26 batch, is pitching idle enterprise GPU capacity as an untapped resource. The claims are bold; the verification is still pending.
Strong earnings give HPE cover to declare victory on its biggest-ever acquisition — but the real test is whether the integration holds as AI infrastructure spending matures.
After years of delays, Valve has named a season. What the company's new Verified programs tell us — and don't tell us — about the hardware.
A basic misconfiguration turned a standard directory service into a plaintext credential vault. The lesson isn't new, but the failure keeps recurring.
The newest Qwen model is cheaper and more capable than its predecessor. It's also proprietary, which matters more than the benchmark numbers.
The Google parent says enterprise and consumer appetite for its AI products is outpacing available infrastructure — a supply constraint that an $80 billion capital raise is meant to fix.
AWS is pitching serverless OpenSearch as the natural fit for agent-driven workloads — but the system depends on a proprietary storage layer, and the gap between the pitch and the plumbing is worth understanding.
A new 7700X3D debuts at $329 while the five-year-old 5800X3D gets a second life at $349. AMD is betting platform longevity is a selling point. The data suggests it might be right.
The decentralized social network is adding long-form publishing — a direct shot at X's Articles feature and the broader creator economy.
Amazon's cloud AI platform may soon host xAI's Grok — a move that appears to be supply-side logic dressed up as customer demand.
Running AI inference closer to the data, inside the database itself, could cut the energy and sovereignty costs of centralised cloud AI. The case is plausible. The evidence is still thin.
The AI chip startup best known for its blazing-fast inference speeds is reportedly shifting its strategic focus — just as Nvidia made a $20 billion move that didn't quite add up to an acquisition.
Law enforcement took down one of the largest botnets on record. The infrastructure was allegedly used to sell anonymized internet traffic through a residential proxy network.
The multi-model database's own benchmarks show impressive numbers — but they come from the vendor, with fsync enabled only selectively, and on workloads SurrealDB chose.
The independent C17 compiler built outside the GCC and LLVM ecosystems is closing its public chapter. What that means for the developers who relied on it.
New benchmarks suggest the hardware bar for running large language models locally may be lower than the industry has been telling you.
At a show defined by next-gen launches and eye-watering prices, AMD is telling desktop PC gamers that their existing hardware is fine — and promising to keep it that way until 2029.
The logistics startup wants to be the infrastructure layer for brands that want Prime-like delivery without handing their customer relationships to Amazon.
Years after its release, the VideoLAN-backed dav2d library underpins video playback across browsers, devices, and streaming platforms — with little public fanfare.
A headline number masks a persistent access problem that no single strong quarter can fix.
Deflock's milestone reveals just how dense the ALPR surveillance grid has become — and how little of it the public could see before volunteers started counting.
PrismML's Bonsai Image 4B uses extreme weight quantization to shrink a large generative model onto consumer hardware. The approach is genuinely novel. The performance claims need scrutiny.
Turnstile, Cloudflare's answer to invasive CAPTCHAs, silently depends on WebGL — a graphics API that leaks enough hardware detail to fingerprint most devices.
A wave of first-generation agentic deployments is failing in production. The culprit isn't the models. It's the infrastructure underneath them.
The consulting giant's acquisition of the world's most-used internet speed measurement platform signals a push to sell network performance data directly to enterprise clients.
Digital Equipment Corporation's DECmate II ran a 12-bit PDP-8 instruction set inside a desktop machine sold as an office word processor — a design decision that was already a decade old when the product shipped.
The columnar database company's revenue growth signals a maturing market for OLAP infrastructure — and a public offering may be closer than the industry expected.
The startup claims its electronically controlled lenses can shift focus in real time. The hardware is promising. The unknowns are significant.
The consulting giant's acquisition of Ookla's consumer-facing tools hands a Fortune 500 firm a trove of real-time internet performance and outage data from millions of users worldwide.
Europe wants control over its own data infrastructure, but dependency on US hyperscalers runs deeper than any regulation can quickly fix.
Tanker diversions and insurance surcharges triggered by the Strait of Hormuz crisis are now pushing up container shipping costs — a supply-chain shock with direct implications for tech hardware imports.
A deep-dive reverse-engineering of the 1980 floating-point coprocessor reveals microcode structures that Intel never publicly documented, raising fresh questions about what else lies buried in legacy silicon.
A setback for Jeff Bezos's rocket company, a new U.S. military launch site in the works, and China's Tiangong station just got a long-stay astronaut — here's what actually matters in this week's launch roundup.
The API aggregator that lets developers swap between dozens of AI models with a single integration just secured a nine-figure round — a bet that no single model will dominate.