Artificial intelligence is rapidly moving from a research curiosity into the center of investment workflows, but two conversations are now dominating: whether investors can trust AI-driven strategies with real capital, and whether Big Tech's enormous AI infrastructure spending will produce acceptable returns.
InvestorAI CEO Bruce Keith, speaking on Zero Sum, argued that general-purpose large language models are useful for explaining market moves and summarizing portfolios, but poorly suited to forecasting breakouts. His firm uses computer vision technology — the same approach behind facial recognition — to turn market data into visual patterns that text-based models cannot detect. Keith said AI is strongest in short-term decisions of roughly two to three months, while humans remain better at longer-term judgment. He also emphasized that investors should treat AI tools like human fund managers, requiring win rate, beat rate, average returns and drawdowns rather than relying on a single backtest or headline performance number.
At the same time, AI infrastructure spending by major technology companies is under increasing scrutiny. Estimates cited by Coinpaper put 2026 capital expenditure by five U.S. hyperscalers at about $697 billion, with capex consuming 93% of operating cash flow, up from 33% in 2023. Microsoft reported June-quarter capex of roughly $41 billion, with Azure and cloud-services revenue up 43% and Microsoft Cloud revenue reaching $59.3 billion; Microsoft 365 Copilot surpassed 30 million paid seats. Alphabet's Google Cloud revenue jumped 82% to $24.8 billion in Q2 2026, while its first-half capex reached $80.6 billion and full-year capex guidance was raised to $195–$205 billion.
Meta's revenue rose 28% to $60.8 billion, but quarterly free cash flow collapsed to $784 million as capex hit $31.1 billion. Amazon showed a similar pattern: AWS revenue grew 37% to $42.2 billion and operating income reached $16.6 billion, but trailing-12-month free cash flow turned negative at -$7.6 billion. Nvidia provided a demand-side signal with data-center revenue of $89 billion, up 117% year over year, and total quarterly revenue of $96.2 billion. Investors are being urged to track whether AI-linked revenue is accelerating, whether free cash flow stabilizes, margins improve, utilization stays high, and whether return on invested capital exceeds the cost of capital — rather than assuming all AI spending is automatically profitable or wasteful.
