Alphabet (NASDAQ:GOOGL) DeepMind executive Jasjeet Sekhon says the technology industry's unprecedented infrastructure spending is laying the foundation for recursive self-improvement, or RSI, in which artificial-intelligence systems help create increasingly capable successors. The argument supports the long-term AI bull case, but Sekhon also acknowledged the immediate problem confronting investors: current revenue does not justify the scale of investment.

RSI would represent a major shift from companies manually training each new model generation. An AI system capable of improving its own architecture, training methods, or research process could accelerate development and make today's data-center buildout considerably more valuable.

Sekhon said AI revenues don't sustain the capital expenditures we're making so far, according to The Information. However, he argued that betting against the industry's technological progress would appear to be unwise, adding that steam engines were used to create the next steam engine.

Alphabet is making that wager on an extraordinary scale. The company raised its 2026 capital-expenditure forecast to between $195 billion and $205 billion from $180 billion to $190 billion, primarily to accelerate the capacity needed to meet customer demand. Management also expects spending to rise significantly again in 2027.

There is already evidence that the infrastructure is generating business. Second-quarter Google Cloud revenue surged 82% to $24.8 billion. However, Alphabet recorded negative free cash flow of roughly $5.9 billion as investment overwhelmed operating cash generation, sharpening concerns that spending and monetization are moving at very different speeds.

Investor takeaway

RSI remains a technological possibility rather than a forecastable earnings catalyst. Alphabet investors should monitor Cloud growth, backlog conversion, AI-product revenue, depreciation and free cash flow for evidence that today's spending is producing acceptable returns.

For Nvidia, Broadcom and other infrastructure stocks, continued hyperscaler spending supports near-term demand. The risk for Alphabet, Amazon, Microsoft and Meta is Sekhon's potential AI air pocket such that the expenditures happen but the revenues don't show up. That scenario would pressure margins, cash flow and AI-stock valuations well before RSI delivers any economic payoff.