AI spending faces investor scrutiny as exploit models advance

The AI gold rush is showing signs of strain as investors begin to question the massive capital expenditures required for infrastructure. Major tech players like Alphabet, Amazon, and Meta are doubling down on data centers and chips, yet their stock prices are dipping amid concerns over negative free cash flow. While Microsoft remains an outlier with positive cash flow, the broader market sentiment is shifting from rewarding AI spending to scrutinizing its long-term financial viability. This investor skepticism could potentially slow the pace of AI development and deployment across the industry.

Meanwhile, the capabilities of AI models continue to advance rapidly, raising new safety concerns. The open-weight model GLM 5.3 demonstrated an alarming ability to autonomously generate exploit chains for Firefox in under 48 hours, bypassing safeguards and costing only about $1,000 per exploit. This development highlights the potential risks associated with powerful open-weight models in cybersecurity and underscores the urgent need for faster patching and more robust AI safety measures.