Discover your next thesis.
Find the businesses behind the trends. Follow the spending.Common questions
Which companies actually collect the money from AI data-centre spending?
Far more than the chip designers, which is the point of looking at it as a flow. A dollar of data-centre capital expenditure is split across accelerators, the memory stacked beside them, networking and optics, the contract manufacturers assembling the racks, and the power and cooling equipment without which none of it runs. Each flow here lists the companies positioned to receive the spending, an estimate of revenue per $1,000 spent, and a "what to watch" note on what would break the thesis.
How are the revenue-per-$1,000 estimates produced?
They are hand-maintained rounded annual estimates, labelled with the year they describe, and they are orientation rather than measurement. Estimates can overlap, because one company's revenue is another's cost, which is why the listed amounts can add up to more than the $1,000 being split. The revenue growth shown next to each company is live and company-wide, year over year — it is not the company's growth from this flow, and the page says so.
Is this a stock screener?
No. It lists up to ten companies of $2 billion market cap or more for each of six sectors and four curated themes, with revenue and market cap, as a place to start reading. There is no filtering by valuation, no custom screens, no backtesting and no performance history. It is built to answer "who is in this business" rather than "which of these should I buy".