Editor’s Note
Chipmakers described demand they cannot meet, and AI companies laid out enormous spending plans. But the week also brought a closer look at what those companies actually earn, and a sharp reminder that higher fuel and borrowing costs can reach company profits. The central question is whether AI revenue will catch up with the investment.
AI Chip Demand Is Running Ahead of Supply
AMD CEO Lisa Su said customers want more chips than AMD can make. The company plans to “substantially increase” supply in 2027, while memory remains broadly supply constrained. Micron’s record quarter put a number on that demand: its fiscal Q4 2026 revenue was about $54.2 billion, up 380% from a year earlier, and gross margin reached 87% of sales. Core data center revenue was about $18 billion, compared with about $1.6 billion a year earlier.
Other suppliers are describing a much larger market ahead. Marvell raised its FY28 revenue outlook to about $20 billion from $18 billion. Its CEO sees a roughly $400 billion AI market by 2030, against the company’s earlier $94 billion opportunity estimate for 2028. Qualcomm’s CFO described an Amazon transaction as a “10-year $60 billion” agreement, with only a portion committed, and set data center revenue targets of $5 billion next year and $15 billion in 2029.
These are strong demand and growth claims, but some describe plans or targets, not results already delivered. That distinction matters when the industry’s investment needs are also climbing. A supplied video cited $1.5 trillion in hyperscaler spending commitments, plus $900 billion in leases. Broadcom’s role arranging more than $50 billion in financing for an OpenAI custom chip was reported as conditional, “if confirmed.”
AI Revenue Is Growing, but the Measures Differ
OpenAI’s annualized revenue figure came under closer scrutiny. CNBC confirmed a roughly $50 billion run rate at the end of September, below the $68 billion figure widely reported last month. A person familiar with the matter said the larger number included gross revenue from partners. OpenAI’s investor presentation also reported 77% total run rate growth in the third quarter and 107% growth for its enterprise business.
That is fast growth, but it does not settle the return question. OpenAI faces pressure to justify an $852 billion valuation. It closed a $122 billion funding round in March and is discussing another round that could raise around $30 billion, though no term sheet has been finalized. Anthropic told investors its annualized revenue run rate reached $65 billion at the end of July. It is reportedly seeking a $2 trillion valuation, while Reuters, citing a leaked prospectus, reported $4.6 billion in 2025 revenue and a $42 billion net loss.
The week also showed AI moving toward consumer products. Meta introduced a personal agent with a free tier and plans costing $20 or $100 a month. Meta says it can book appointments and fill out forms, and is exploring a share of shopping transactions, but Wang said there is no concrete plan. Users must opt out if they do not want their interactions used to train Meta’s models. The product gives Meta a possible way to bring AI to users directly, while the company is still under pressure to show returns on its AI investments.
Oil and Yields Are Testing Company Profits
Weak hiring lowered traders’ odds of an October Fed hike, but bond yields stayed high. September payrolls rose 29,000, against expectations of roughly 84,000 to 90,000, and unemployment reached 4.2%. The 10-year Treasury yield was reported above 5.35%, its highest in 24 years, before other reports described yields falling across the curve. Stocks recovered, and the S&P 500 and Nasdaq 100 recorded their highest weekly closes on record.
Oil added a separate cost and inflation concern. Brent traded above $100 on Thursday and remained near $103 to $104 on Friday. Delta cut its 2026 adjusted profit forecast to roughly $5.10 to $5.60 a share, from $6.50 to $7.50, and expects fuel costs to rise by about $6 billion. Its third-quarter fuel expense jumped 62%. Fed Governor Christopher Waller said more rate hikes are likely needed to bring inflation back to 2%, though the timing is flexible. Fuel costs are already weighing on one company’s profit outlook, while inflation and rate policy remain unsettled.
Counter-Thesis and Risk Watch
Several supplied voices argued that AI spending and valuations have moved ahead of proven returns. @geopoliticaleconomyreport called the AI boom a bubble and cited $600 billion in 2026 AI-related spending by Microsoft, Meta, Alphabet and Amazon. Ray Dalio called AI a “classic bubble” that may be close to bursting. Separately, @wallstreetmillennial’s account of a Harvard study said engineers produced about 4,000 more lines of code per month twelve months after adopting agentic AI, but the increase in resolved Jira issues was not statistically significant.
Bond and debt concerns also remain. Dalio warned China and Japan could buy fewer U.S. Treasuries, citing China’s holdings falling from $1.3 trillion in 2013 to about $618 billion, while Japan holds roughly $1.1 trillion. A Financial Times account said the bond selloff was prompting companies to rethink borrowing plans and raised the prospect of defaults among the lowest-rated borrowers. These are warnings, not settled outcomes.
Finally, @geopoliticaleconomyreport warned that disruptions around Bab al-Mandab, Hormuz and the Suez Canal could affect oil supply. The speaker said those routes account for 37% of globally traded oil supply. That risk sits alongside the fuel costs and inflation pressure already visible in the period.
Looking Ahead
Can AI companies turn their rising revenue into returns that support the scale of spending and valuations? And do oil prices and bond yields ease enough to relieve pressure on costs and inflation?