Welcome to this week’s edition of The AI Loop, your
weekly capsule of high-impact AI news. From big infrastructure bets to new partnerships and national
strategy moves, it’s been a big week in the space. Grab a minute and catch up with us!
Ambassador Program Applications Open
We’ve opened applications for the first wave of Perlix AI ambassadors. This will be a small group of committed contributors who want to help shape how the Human Layer for AI grows across regions and communities.
As an ambassador, not only will you help bring local communities into the network, you’ll expand your personal network, strengthen your voice in the Web3 and AI space, and be recognized for the impact you create as the ecosystem scales.
OpenAI spreads $600B cloud AI bet across AWS, Oracle, Microsoft
OpenAI is expanding its $600B cloud investment across multiple providers, securing access to the NVIDIA GPUs and custom infrastructure needed to train and run frontier-scale models. The company has reportedly committed about $250B to Microsoft, $300B to Oracle, and $38B to AWS as part of a long-term multi-cloud strategy.
This shift highlights a new reality. High-performance compute is no longer something companies can access on demand. It must be reserved and built out years in advance. Even with the new AWS agreement, the full capacity won’t be online until late 2026.
The future of AI is now shaped as much by hardware timelines and supply chain strategy as it is by model breakthroughs.
Source: AI News
Apple nears $1 billion-a year deal to use Google AI for Siri
Apple is stepping up in the AI assistant race. The company announced plans to use a custom version of Google’s Gemini model to power a new Siri experience as early as spring 2026, with the partnership valued at roughly $1 billion per year.
The setup will be hybrid. Apple’s on-device models will continue handling personal and privacy-sensitive queries, while the custom Gemini model will run in Apple’s private cloud to handle deeper reasoning and large-scale tasks.
The takeaway: even the largest platforms are choosing to borrow model capacity while they build their own. Speed to market now matters as much as owning the entire stack.
Source: Bloomberg
South Korea to deploy 260,000 NVIDIA GPUs for sovereign AI
At the APEC Summit, NVIDIA CEO Jensen Huang announced a national initiative with the South Korean government and major companies including Samsung, SK Group, Hyundai, LG, and NAVER. The plan calls for deploying more than 260,000 NVIDIA GPUs across sovereign cloud infrastructure and industrial AI “factories” to support robotics, automotive systems, manufacturing, and research.
This marks one of the largest country-level AI investments to date and highlights a growing global trend. AI infrastructure is beginning to be treated as strategic national infrastructure, similar to power grids and semiconductor fabs. Sovereign AI is becoming a matter of economic resilience and future competitiveness.
Source: Reuters
Quick Bites
OpenAI introduces Aardvark for automated security research
OpenAI launched Aardvark, a GPT-5 powered system that autonomously identifies and patches software vulnerabilities across enterprise and open-source codebases. It uses deep code reasoning, commit-aware monitoring, and sandbox validation, and is currently in private beta.
Source: Open AI
UMG and Stability AI team up to build licensed AI music tools
Universal Music Group and Stability AI are collaborating on next-generation AI music creation tools trained on licensed data with artist participation. The goal is to enable creative experimentation while maintaining copyright and commercial safety standards. This points toward a future where AI music tools respect both artists and rights holders.
Source: Stability.AI
Stay in the Loop
That’s a wrap for the first edition of The AI Loop. Each week, we’ll track the breakthroughs, power plays, and shifts shaping the world of AI.
Join the Perlix AI community on Telegram to stay connected and catch the next drop of insights from across the ecosystem.