Palantir grew revenue 93% while generating a 63% adjusted free cash flow margin. At least one company is converting AI demand into realized economics today.
The broader question is no longer whether AI demand exists. It is where that demand becomes durable cash flow, and where returns still depend on scarcity, leverage, and uncertain future economics.
Palantir Is Converting AI Demand Into Cash Now
Palantir reported Q2 2026 revenue of $1.94 billion, above the $1.8 billion expectation discussed before earnings.
Growth was broad. U.S. commercial revenue rose 149% to $764 million, while U.S. government revenue increased 90% to $809 million.
More importantly, that growth produced current profitability and cash generation. GAAP net income reached $1.06 billion, a 55% margin. Adjusted free cash flow was $1.22 billion, a 63% margin.
The results support the thesis that enterprise AI value may accrue at the application layer, where software connects with customer operations and creates measurable value. They do not prove that every application software company will win.
Palantir’s distinction is that investors are not being asked to wait until 2030 to see whether demand appears. Revenue, margins, and cash flow are appearing now.
Memory Scarcity Signals Demand, Not Returns
The three major memory suppliers have reportedly pre sold all available DRAM and high bandwidth memory capacity for 2027. Suppliers can reportedly satisfy only 60% to 70% of requested capacity, with cloud service providers and AI companies receiving priority over smartphone and PC manufacturers.
If accurate, that suggests AI infrastructure demand is reshaping capacity allocation across adjacent industries. It may also improve demand visibility for memory suppliers.
But there is no company specific allocation data, contract pricing, supplier level financial forecast, or evidence showing how much of the shortage is already reflected in expectations.
The demand signal is meaningful. The investment conclusion is incomplete.
Oracle’s Backlog Comes With Financing Risk
Oracle reported 93% cloud infrastructure growth, 119% growth in infrastructure CPU and GPU activity, and $648 billion in remaining performance obligations.
Oracle also trades at a P/E ratio of about 20, while analyst targets imply roughly 90% upside. The bullish case is that the market may be underestimating the eventual economics of its infrastructure buildout.
The financing side complicates that case. Projected cash outlays are rising from $50 billion to $70 billion, while debt and preferred share issuance are also increasing.
The unresolved questions include demand in 2030, future capacity pricing, customer profitability, and the economics of today’s chips compared with those available in 2030. Capacity can be scarce while the investors funding it still earn inadequate returns.
Expensive Capital Raises the Standard
Carlyle CEO Harvey Schwarz expects inflation to remain stickier and the cost of capital to stay higher. He reported 9% to 10% EBITDA growth across Carlyle’s portfolio companies and continued economic resilience, alongside early signs of consumer weakness.
He also acknowledged stress in SaaS lending. Loans made in 2021, 2022, and 2023 now face uncertainty around large language models, AI, and software multiples.
A durable technology shift can coexist with poor loans, excessive leverage, and disappointing asset returns. AI demand can be real without making every supplier, lender, or infrastructure builder an attractive investment.
The cleaner signal is who converts AI demand into cash before financing assumptions become the investment thesis.