1. University of Pennsylvania Breakthrough Photon-Enhanced AI Chip Cuts Power Use by 90%
Researchers at the University of Pennsylvania have unveiled a new hybrid light-matter exciton-polariton technology that can drastically accelerate AI computing while slashing energy consumption. The innovation combines photons and electrons in atomically thin semiconductors to enable all-light signal switching, eliminating the energy-heavy conversion between optical and electronic signals that drains power in traditional AI hardware. The team demonstrated that the new system can complete a full computing switching operation using only 4 quadrillionths of a joule of energy. Once fully scaled, the photonic chips will be able to process raw data directly from cameras and sensors, and are projected to reduce the total power demand of large-scale AI data centers by more than 90%.
2. OpenAI Partners with Getty Images to Build Fully Licensed AI Visual Content Ecosystem
Global leading visual content provider Getty Images announced a formal content partnership with OpenAI on June 22, bringing its entire library of legally licensed professional images into the ChatGPT ecosystem. Under the agreement, Getty’s copyrighted visual resources will be directly accessible in ChatGPT’s search and generation workflows, eliminating long-standing copyright disputes over AI-generated visual content. This deal follows OpenAI’s earlier collaborations with Adobe, Microsoft and NVIDIA, marking a major milestone in the global push to standardize compliant, legally protected AI content generation.
3. Global Weekly AI Large Model Total Calls Hit 46.7 Trillion Tokens, 9 Straight Weeks of Growth
Latest data from OpenRouter shows that between June 15 and 21, 2026, the total global call volume for AI large models reached 46.7 trillion tokens, marking a 4.7% week-on-week increase and the 9th consecutive week of rising demand. Industry analysts point out that the dual drivers of booming consumer AI application adoption and large-scale expansion of global AI infrastructure are pushing the sector into a new high-growth phase, with optical interconnection emerging as the fastest-growing core track across the entire AI hardware supply chain.
4. Ubtech’s U-World Consumer Humanoid Robot Exceeds 5000 Pre-orders on Launch Day
Chinese robotics leader Ubtech announced on the evening of June 21 that its new full-size ultra-realistic consumer humanoid robot under the brand “U-World” has accumulated more than 5000 pre-orders within hours of its debut on JD.com, with simultaneous sales now live on Tmall. Industry reports note that 2026 is the critical inflection point for the humanoid robot market to move from 0 to 1, with global players on track to expand total annual production volumes from thousands to tens of thousands of units in the second half of the year.
5. Tencent’s WeChat Native AI Assistant “Xiaowei” Enters Small-Scale Gray Testing
WeChat has launched limited internal testing for its built-in native AI assistant “Xiaowei”, with a small group of users now seeing a new eye-shaped test entry icon in the top-left corner of their main WeChat interface. The current version supports one-tap generation of custom mini-programs through natural language prompts, though the generated mini-programs are only accessible to the creator for now. Analysts estimate that the feature will unlock massive new UGC potential across WeChat’s 1 billion+ user base, and drive a sharp surge in Tencent’s AI computing demand once fully rolled out.
6. DeepSeek Completes 51 Billion RMB Historic Financing, Founder Holds Full Operational Control
Chinese AI startup DeepSeek has closed the largest single financing round in the country’s AI history, raising over 51 billion RMB in its first public funding. Founder Liang Wenfeng personally invested 20 billion RMB, with Tencent contributing 10 billion RMB, CATL putting in 5 billion RMB, and NetEase, JD and IDG Capital each investing 3 billion RMB. The unique deal structure grants no voting rights or board seats to institutional investors, and all shares are locked for 5 years to guarantee long-term independent technical development. The company’s post-money valuation now stands close to 400 billion RMB.
7. Google DeepMind Releases AlphaFold 3 for Free Academic Access
Google DeepMind has officially opened full free access to AlphaFold 3 for all non-commercial academic researchers worldwide. The new model delivers unprecedented accuracy in predicting complex protein structures and molecular interactions, and is projected to cut years off the typical multi-year drug discovery timeline. Major global pharmaceutical companies have already begun integrating the tool directly into their core R&D pipelines to accelerate new treatment development.
8. Critical Gravity SMTP WordPress Plugin Vulnerability Exposes Thousands of Sites to API Key Theft
A zero-day vulnerability in the widely used Gravity SMTP WordPress plugin is being actively exploited by global hackers, who are targeting websites to steal exposed API keys for Mailjet, Zoho and other mainstream email services. The flaw allows attackers to remotely execute arbitrary code and access backend system credentials without any user interaction, with more than 6000 sites already confirmed compromised as of June 21. Cybersecurity teams have released urgent patches and advised all WordPress administrators to update the plugin immediately to prevent data loss.
9. OpenAI Launches Life Sciences Research Benchmark, Top Model Fails 63.9% of Real-World Tests
OpenAI officially released LifeSciBench, a dedicated benchmark suite with 750 practical tasks designed to evaluate whether AI systems can support real life science research, rather than just answer basic biology quiz questions. The company’s highest-performing specialized model GPT-Rosalind took the top spot in the rankings, but still failed 63.9% of the test scenarios, exposing major gaps between current AI capabilities and the strict, error-sensitive demands of experimental design and drug development.
10. Global AI Agent Market Surges 120% Month-on-Month as Enterprise Adoption Accelerates
Latest market tracking data shows that the global AI agent market recorded a 120% month-on-month growth in June 2026, with more than 30% of Fortune 500 companies now deploying custom AI agents for customer service, internal workflow automation and supply chain management. The rapid adoption is driving new demand for specialized agent frameworks and lightweight inference chips, creating a whole new fast-growing segment within the broader AI industry.