A company can fund a new data centre with money it earns, money it borrows or money from new shareholders. Each route has a cost. Borrowing brings interest and repayment obligations. Selling shares spreads ownership across more investors. That can make the company financially stronger while leaving each existing shareholder with a smaller claim on future profits—a process called dilution.

Meta made that trade-off tangible on Monday. It launched about $35 billion of offerings, combining common shares with preferred securities that must later convert into shares. It also outlined a possible future share-sale programme of about $60 billion; that programme is not established and the proceeds are not cash already received. Meta reports substantial cash resources and intends to use several funding sources. This is evidence of the scale of its investment ambition, not proof that the business is running out of money.

Anthropic’s IPO filing brings a different part of the same story into view. An initial public offering lets investors buy into a business previously financed privately. Anthropic has not determined the offering’s share count or price range. Until those terms arrive, enthusiasm about the technology cannot settle whether the shares offer value. More capital could support more computing purchases, benefiting suppliers, while the return to new investors still depends on the price paid and the profits eventually available to them.

Monday’s market welcomed the prospect: the Nasdaq gained 0.82%, ahead of the S&P 500’s 0.70%. Oil also eased, providing some relief from last week’s pressure. But the US Central Bank (Fed) decision and major technology earnings are still ahead. A rally before those tests does not establish that financing has become cheap or that every planned project will earn its cost. For suppliers, orders backed by customers with dependable funding deserve more confidence than expansion plans that require another fundraising round.

We continue to favour selected AI infrastructure, but financing terms belong beside demand and revenue in the investment case. A useful test is whether profit per share can grow after allowing for the new shares issued and the full cost of serving customers. For New Zealand investors holding global technology funds, the same distinction applies: stronger industry growth does not automatically mean stronger returns from every company in the fund.