From Nvidia’s grand imaginative and prescient of trillion-dollar AI factories to new cracks in cell and enterprise safety, this week in tech made one factor clear: the race to construct smarter methods is accelerating quick, and so are the dangers, prices, and penalties that include them. Throughout AI, infrastructure, cybersecurity, and company technique, firms are pushing deeper into an period wherein brokers, automation, and large compute bets are reshaping how tech is constructed, secured, and offered.
High information
Nvidia’s increasing AI empire
Nvidia continued its dominance at GTC 2026, rebranding knowledge facilities as trillion-dollar “AI token factories.” CEO Jensen Huang projected a $1 trillion income alternative by 2027, unveiling the Vera Rubin platform — a seven-chip stack that enhances inference throughput by as much as 35× per megawatt — and launching the Nemotron Coalition for enterprise AI brokers.
In a complementary keynote, Huang described the rise of a “token-driven AI economic system,” the place inference workloads dominate and knowledge facilities evolve into AI factories. Nvidia is constructing infrastructure and frameworks like NemoClaw to guide this subsequent part of computing.
In the meantime, Nvidia additionally introduced the Area-1 Vera Rubin Module and different radiation-hardened GPUs for orbital knowledge facilities. Companions like Starcloud and Planet Labs plan to make use of these chips to course of knowledge instantly in orbit, marking a brand new frontier for space-based AI computing.
And in robotics, Nvidia unveiled new simulation instruments and robotic brains, together with Cosmos 3 world fashions, Isaac simulation frameworks, and the GR00T N1.7. Industrial companies are adopting its Omniverse and Isaac Lab platforms, whereas cloud suppliers combine its Bodily AI Information Manufacturing facility to generate artificial coaching knowledge.
AI and app ecosystem shifts
Xiaomi revealed that the mysterious “Hunter Alpha” AI mannequin on OpenRouter was really its MiMo-V2-Professional system — a trillion-parameter mannequin launched stealthily in March. The system demonstrated huge processing energy with low-cost inference and could also be open-sourced later, sending Xiaomi’s shares upward.
Apple paused updates for AI app builders Replit and Vibecode, citing App Retailer guidelines that prohibit apps from executing new code post-review. The corporate is working with builders to carry their apps into compliance after a surge in AI-assisted submissions.
Instagram introduced it’s going to finish end-to-end encrypted chats on Could 8 attributable to low utilization, restoring Meta’s skill to scan messages for moderation and legislation enforcement. Customers are inspired to export chat histories or swap to WhatsApp or Sign for personal messaging.
Google and Anthropic increase AI options
Google expanded its Private Intelligence function to US customers, permitting opt-in connections throughout Gmail, Pictures, and YouTube for customized help. The system emphasizes privateness with encrypted knowledge and non-compulsory app connections.
Google Maps additionally obtained its largest replace in a decade, including an “Ask Maps” chat function powered by Gemini for customized suggestions and bookings, plus an immersive 3D navigation mode rolling out throughout platforms.
Anthropic launched interactive charts and diagrams inside Claude, enabling customers to visualise knowledge in actual time and enhancing conversational move in comparison with ChatGPT and Gemini.
Infrastructure and automation
Meta signed a $27 billion, five-year cope with Amsterdam-based Nebius for GPU cloud capability, increasing its AI infrastructure and boosting Nebius’s profile as a “neocloud” supplier. The deal may push international AI infrastructure spending towards $700 billion this 12 months.
Boston Dynamics and Ghost Robotics are deploying robotic canine to protect knowledge facilities, utilizing sensors to detect warmth, leaks, and intrusions. Operators like Novva Information Facilities report improved reliability and price financial savings over human patrols.
Insider intel
The Neuron argues that the standard app period is ending, giving option to clever brokers that autonomously carry out duties and handle workflows — marking a serious shift in how customers work together with know-how.
A joint examine by GovAI and Brookings, lined by eWeek, discovered that 37.1 million US jobs are extremely uncovered to AI, although 26.5 million employees have the adaptability to retrain. Clerical and administrative roles — 86% held by ladies — face the best threat, prompting requires focused upskilling to scale back gender and regional disparities.
Safety alerts
Main exploits and vulnerabilities
DarkSword, a brand new iOS exploit, permits attackers to hijack iPhones through malicious webpages, stealing knowledge and crypto wallets. Linked to Russian espionage teams, it impacts iOS 18.4–18.7 and has been patched, although many customers stay weak.
MediaTek confirmed a flaw (CVE-2026-20435) in its safe boot chain that permits attackers with bodily entry to extract PINs and encryption keys from Android telephones in beneath a minute. The difficulty impacts roughly 875 million gadgets.
Microsoft patched CVE-2026-26123, a flaw in its Authenticator app that might let malicious apps hijack one-time codes or login hyperlinks. Customers are urged to replace instantly.
Microsoft additionally launched an out-of-band hotpatch (KB5084597) to repair three important RCE flaws in Home windows 11’s Routing and Distant Entry Service (RRAS), tracked as CVE-2026-25172, -25173, and -26111.
Company and infrastructure assaults
CISA issued an alert after pro-Iran hackers used Microsoft Intune to wipe almost 80,000 Stryker gadgets. The company advises implementing least-privilege entry and enabling multi-admin approvals.
Stryker confirmed the identical assault disrupted international operations, although no ransomware was detected. Restoration efforts are ongoing.
Intuitive Surgical disclosed a phishing breach that uncovered worker and buyer contact knowledge however spared affected person and scientific methods because of community segmentation.
Software program and AI safety fixes
Apple launched Background Safety Enhancements (BSIs) to ship silent Safari patches between main updates, beginning with a WebKit repair for cross-tab knowledge leaks.
Microsoft Copilot was discovered weak to cross-prompt injection (CVE-2026-26133), permitting attackers to insert pretend safety alerts or phishing hyperlinks into AI-generated summaries. The difficulty was patched on March 11.
Business shakeups
Acquisitions and AI investments
OpenAI introduced the acquisition of Astral, the developer behind Python instruments uv, Ruff, and ty, to strengthen its Codex AI coding assistant. The deal goals to increase Codex’s scope past code technology into full venture administration and evaluation.
Layoffs and restructuring
Meta is reportedly contemplating chopping as much as 20% of its workforce to fund AI growth, following huge infrastructure investments and comparable strikes throughout the tech sector.
Atlassian introduced a ten% workforce discount — about 1,600 jobs — to redirect assets towards AI and enterprise gross sales initiatives, mirroring comparable cost-cutting methods at Oracle, Salesforce, and Meta.
If you wish to see extra from our publication, try the Day by day Tech Insider archive.
