The artificial intelligence economy is experiencing unprecedented growth, but beneath the glittering surface of new models and billion-dollar investments, structural cracks are emerging that could slow the race. At the recent Milken Global Conference in Beverly Hills, five key players from the AI supply chain analyzed the critical pain points, touching on everything from chip shortages to the possibility that the entire architecture underpinning the technology is flawed. Adding to these voices, new events paint a complex picture for the future of tech.
Energy and Data Centers: Microsoft's Paradox and the Rise of Neoclouds
Microsoft has announced ambitious expansion plans for its AI data centers, but these very plans are coming into direct conflict with the company's clean energy goals. The push for new computing facilities demands massive amounts of electricity, jeopardizing promises of carbon neutrality. This paradox is not isolated. According to insights from the conference, AI energy demand could outpace current grid capacity, forcing companies to rethink their strategies. Meanwhile, Elon Musk's xAI is quietly shifting its business model. No longer solely a builder of AI models, xAI is transforming into a neocloud company selling computing infrastructure and data centers instead of focusing exclusively on training language models. This shift reflects a broader industry trend: the true economic value may lie not in AI itself but in offering the computational power needed to run it. Our article on SpaceX's Terafab chip factory shows how semiconductor production is becoming a key pillar of this infrastructure.
Privacy Under Fire: Canadian Allegations Against OpenAI
Canadian authorities have accused OpenAI of violating federal and provincial privacy laws, raising concerns about personal data collection and consent practices. This is no minor issue. As generative AI spreads, the tension between innovation and data protection is becoming sharper. If regulations tighten penalties, the cost of developing and deploying language models could rise significantly. The Canadian case fits into a global context of increasing regulatory scrutiny that could redefine the rules for the entire AI ecosystem.
Google's AI Search: When Reddit Becomes an Expert
Google has announced a significant update to its AI Mode and AI Overviews. The most controversial feature is the introduction of an Expert Advice section, which draws heavily from Reddit and social media. AI responses will now include insights labeled as expert advice, even though they originate from public online discussions. Google emphasizes that the section might also be called Community Perspectives depending on the query, and that the creator's name or community will always be shown for transparency. Additionally, Google introduced further exploration suggestions, Subscribed labels for links from subscribed news sites, and a new link preview on hover for desktop users to reduce hesitation. This move raises deep questions about source credibility and information quality. If Google's AI learns from forums like Reddit, the risk of spreading misinformation under the guise of authority is real.
Barry Diller's Warning on AGI
Barry Diller, the media mogul, publicly defended OpenAI CEO Sam Altman but issued a stark warning: as AGI (Artificial General Intelligence) approaches, trust becomes irrelevant. Without proper guardrails, the power of AGI could become an unpredictable force. Diller's words echo the concerns of the very architects at the Milken Conference, many of whom believe that the current technical architecture of AI is fundamentally wrong or too fragile to scale without collateral consequences. The combination of energy, regulatory, and credibility problems is creating a perfect storm. Meanwhile, companies like Apple are seeking to diversify chip production (as covered in this related article), but the road to sustainable and ethical AI remains long. The ecosystem will not collapse overnight, but the signs of strain are evident and demand deep reflection from investors, regulators, and developers.
Sponsored Protocol