AI Is Eating the World: $400B Bets, Agentic Breakthroughs & EU Crackdowns
4th Aug 2025 | Superintelligence Newsletter
Hey Superintelligence Fam 👋
The AI industry isn’t just growing—it’s exploding. Big Tech invested $155B in AI infra last year, with $400B projected next year. That’s more than the GDP of over 150 countries combined.
From China's GLM-4.5 to Gemini solving Olympiad problems, and EU’s landmark AI Act now live, we’re entering an era where AI is both unstoppable and under a microscope. Buckle up—it's only the beginning.
Big Tech’s $400B AI Infrastructure Expansion for 2026 : Big tech titans poured $155 billion into AI infrastructure in the past year, eyeing an explosive $400 billion expansion next fiscal year to power next-gen AI and data centers.
GLM-4.5 Open–Source Agentic AI : Chinese AI startup Zhipu has released its open-source GLM-4.5, designed for intelligent agent applications, democratizing advanced AI tools globally amid China’s rapidly evolving AI sector.
EU AI Act Enforcement Begins August 2, 2025. : As of August 2, the EU’s AI Act enforcement enters its second phase, imposing GPAI obligations on major AI providers—mandating transparency, risk assessments, bias testing, and copyright compliance across operations.
Subliminal Learning: Language Models Transmit Behavioral Traits via Hidden Signals in Data : Anthropic et al. reveal subliminal learning: student LMs inherit owl preference and misalignment traits from filtered, semantically unrelated teacher outputs, and MLP achieves >50% MNIST accuracy on noise.
Building and Evaluating Alignment Auditing Agents : Anthropic introduces three autonomous agents—investigator, evaluator, and breadth-first red-teamer—that unearth hidden model objectives, deliver a 42% super-agent solve rate, and surface misaligned behaviors in rigorous audit games.
Gemini 2.5 Pro Capable of Winning Gold at IMO 2025 : Independent researchers leverage a rigorous self-verification pipeline on Gemini 2.5 Pro to solve 5 of 6 IMO 2025 problems, earning a gold-medal score with formal proofs and modular reasoning.
ChatGPT Study Mode : ChatGPT Study Mode transforms AI tutoring with interactive Socratic prompts, scaffolded lessons, quizzes, and personalized feedback—guiding learners step-by-step to build deeper understanding beyond quick answers.
RunwayLM Aleph : Runway Aleph redefines video editing with in-context AI—seamlessly adding, removing, and transforming objects, angles, styles, and lighting on any clip via effortless, interactive text prompts.
Google Deepmind Aeneas : Google DeepMind’s Aeneas resurrects fragmentary Roman inscriptions—predicting missing words, provenance, and dates with 90% historian approval, powered by a 200K-inscription, 16M-character database using incredible accuracy.
Supercharge Your AI Skills
Join our complimentary 7-day email series to explore LLM essentials—from core concepts and model design to training workflows, real-world use cases, and ethical best practices.
Sign up now and take your AI expertise to the next level!
the Indian Institutional Quant Conference spotlighted AI ethics in fintech, China warned AI could become an “exclusive game”, European voice actors pushed back on AI dubbing, Google agreed to join the EU’s AI Code of Practice, and the EU AI Act began enforcement, mandating transparency measures.
Thank you for tuning in to this week's edition of Superintelligence Newsletter! Stay connected for more groundbreaking insights and updates on the latest in AI and superintelligence.
For more in-depth articles and expert perspectives, visit our website | Have feedback? Provide feedback.
If you wish to partner with us then Explore Here
Stay curious, stay informed, and keep pushing the boundaries of what's possible!
Until Next Time!
Superintelligence Team.