30/07/26 Investor’s Daily Brief - AI Is Raising Its Own Cost of Capital AI demand is strengthening growth while making the infrastructure behind it harder to finance. That increasingly favors self-funded platforms over companies asking investors to wait. Market snapshot (equities: July 29 close; WTI/BTC: July 30 morning ET): S&P 500: 7,316.15 (-1.52%) Nasdaq: 24,442.94 (-1.74%) WTI: $83.42 (-1.23%) Bitcoin: $64,763 (+0.30%) The Fed paused. Long-term capital got costlier. The Fed held rates at 3.5%–3.75%, but the 9–3 vote showed a hawkish split. The 30-year yield jumped to 5.228%, its highest since 2007. That steepening reflects inflation, term-premium and funding concerns. For capital-intensive AI, waiting now has a higher price. Microsoft set the cash-conversion benchmark. Azure grew 43%, paid Copilot seats rose from 20 million to 30 million, and revenue climbed 18% to $90 billion. Microsoft stayed free-cash-flow positive while planning more than $50 billion of quarterly investment. Its $678 billion backlog is not all AI, but it offers visibility. Lower reported capex partly reflects longer asset lives and lease classification—not a smaller build. Meta broke the capex patience trade. Revenue grew 28%, yet free cash flow collapsed to $784 million and net income fell 14%. Add $2.4 billion of legal costs, a $130 billion capex floor and a softer third-quarter midpoint, and the 10% after-hours drop makes sense. The ad engine remains powerful; the cash conversion no longer does. Meta’s funding model is now part of the thesis. Meta carries $83.7 billion of long-term borrowings plus large lease commitments and guarantees, while its 40-year bonds yield near 7%. Next year’s analyst capex estimates span $174 billion to $280 billion. That is not a solvency warning. It is a visibility and strategic-flexibility problem. IBM’s quantum milestone remains optionality. IBM may have advanced validation of quantum advantage, but technical progress is not yet an earnings inflection. Independent replication, paid workloads and progress toward fault tolerance matter more than benchmark headlines. Bottom line: This is not an anti-AI market. It is a market becoming less tolerant of unfunded duration. Microsoft shows that huge capex earns patience when adoption, backlog and cash generation move together. Meta shows that even strong revenue cannot indefinitely offset shrinking free cash flow and rising commitments. If long yields stay above 5%, self-financing becomes the AI trade’s decisive separator.