Mark Cuban warns the AI wars could end like the search engine crash — with one winner and a lot of losers (www.businessinsider.com)

Investor Mark Cuban warned that today’s AI arms race — with Google, OpenAI, Meta and others pouring vast sums into ever-larger “foundational” models — looks a lot like the 1990s search-engine scramble and could end the same way: a winner-take-all market that leaves most players bankrupt. He argues companies are overspending on compute, datasets and massive data-center infrastructure in case dominance emerges, creating an economic bubble that could pop if progress stalls or a disruptive technology appears. Cuban believes the real threat is not incremental model scaling but an unforeseen breakthrough that makes current investments obsolete. For the AI/ML community this is a cautionary signal about incentives and fragility: heavy centralization around a few huge models concentrates power, skews research toward scale-heavy approaches, and raises systemic risk for infrastructure builders and startups. Technically, Cuban’s point underscores the importance of algorithmic and hardware efficiency (optimization, new architectures, or novel compute paradigms) that could undercut brute-force scaling. The implication for practitioners and policymakers is to balance scale with innovation in efficiency, open research, and diverse architectures to avoid lock-in and brittle economics if the “AI wars” resolve into a single dominant platform.