Daily Feed

AI

Gemini 3.5 Flash and Gemini Omni ship in Google's May 2026 wave

Google's latest model pair advances agentic reasoning and multimodal creation — now available through the Gemini API and Google AI Studio.

AI

OpenAI readies next-generation GPT-class model roadmap

The company outlined capabilities and model plans for 2026 and beyond, with agentic features and longer context windows as key pillars.

AI Research

New efficiency technique cuts training energy by double-digit percentages

Researchers propose architecture and pipeline changes that reduce energy and compute requirements for frontier-model training runs.

Applied AI

AI assistants enter the smart-home mainstream with on-device routines

Home hub makers bundle local voice, sensors and on-device routines, cutting cloud round-trips for common automation tasks.

Today at a glance

• Gemini 3.5 Flash & Gemini Omni — Google's May 2026 launch pair; advanced agentic reasoning and creation via Gemini API and AI Studio (source)
• OpenAI model roadmap — 2026-plus GPT-class plans with agentic features and extended context windows (source)
• Qwen3-Coder-Next (80B) — Alibaba's open-weight model reaches near-frontier benchmark scores with local-run capability (source)
• Claude Fable 5 — Anthropic flagged in LLM trackers as one of the most-watched 2026 releases (source)

Cloud

Cloud providers extend AI infrastructure deals and cut inference costs

Major providers expand compute clusters and tighten enterprise tiers as demand for inference capacity accelerates into H2 2026.

Cloud

Microsoft commits $17.5B to expand AI and cloud infrastructure in India

Largest-ever Asia investment for Microsoft; spans 2026-2029 with cloud and AI inference capacity targets across India.

Product

Google debuts next-gen TPU at Cloud Next '26

New tensor processing units powering the latest Google Cloud AI services are now live, targeting training and inference workloads.

Chips

MediaTek edge-to-cloud AI strategy on show at Computex 2026

Automotive, connectivity and data-centre chip families span the full stack from end-device inference to cloud training nodes.

Cloud & Infrastructure at a glance

• Google Cloud Next '26 AI infrastructure — new TPU generation, expanded training pods and cost-optimised inference paths now live (source)
• Enterprises invest in agentic AI — up to 75% of enterprises may adopt agentic AI in 2026, driving chip and data-centre spending (source)
• Microsoft India $17.5B commitment — AI and cloud infrastructure expansion through 2029 (source)
• Data-centre AI chip market — valued at $14.87B in 2026, on steep growth trajectory to 2034 (source)

Startups

AI startup funding surges past $33B in 2026 — UK leads Europe

Crunchbase data shows US and UK AI startups already outpacing 2025 totals, with TwelveLabs and Together AI closing nine-figure rounds.

Policy

New US executive order sets AI innovation and security framework

A June 2026 executive order coordinates national AI policy, balancing capability growth with security and national-interest reviews.

Policy

US state AI laws take effect: CO, CA, TX, IL lead the patchwork

Four states now have active AI rules; FTC is actively fining companies — compliance programmes are now a board-level priority.

Security

Prompt injection defences become standard practice in AI deployments

Security teams are standardising layered guardrails as prompt-injection attacks grow in scale, sophistication and enterprise targeting.

Startups, Policy & Security at a glance

• AI startup funding $33B+ — cumulative 2026 figures surpass full-year 2025; UK is the top European hub for AI venture (source)
• Together AI $800M Series C & TwelveLabs $100M Series B — both closed 1 July 2026 (source)
• June 2026 US AI Executive Order — promotes advanced AI innovation while establishing security review requirements (source)
• Four-state US AI law regime active — Colorado, California, Texas and Illinois rules are live; FTC enforcement ramping up (source)

This page and the homepage hero are updated daily by automated news curation.