AI NewsInfrastructureReported

Meta plans to raise 95 billion dollars in equity to pay for AI data centres

Meta said it will raise up to 95 billion dollars in equity. The money pays for its AI data centre build.

AI News

Editorial2 min read

LinkedInX

Why it mattersMeta's own cash flow turned negative last quarter. The equity raise shows what it now costs to keep buying AI compute.

The cost of building AI compute at scale is now written on Meta's balance sheet. Meta Platforms said on Monday it will raise up to 95 billion dollars in equity. The money pays for AI data centres, cloud computing and other infrastructure, according to the company's filing. Business Recorder reports this is the largest equity raise Meta has ever done.

The structure has two parts. One is a 35 billion dollar underwritten public offering, starting now. It splits into about 22.5 billion dollars of Class A common stock. The rest, about 12.5 billion dollars, is depositary shares tied to mandatory convertible preferred stock. The second part is an at-the-market programme of up to 60 billion dollars. This begins once Meta's new non-voting Class C shares start trading in the first quarter of 2027.

Who else is in the deal

Business Recorder reports that three chip makers have told Meta they may buy Class A shares. The combined order would be up to 5 billion dollars. The three are Nvidia, Advanced Micro Devices and Broadcom. These companies make the silicon that runs Meta's AI training and inference. The interest is non-binding.

What this costs Meta right now

Meta spent 2.12 billion dollars more than it took in during the third quarter. Business Recorder notes this was the company's first negative free cash flow in more than a decade. The same quarter one year earlier had positive free cash flow of 10.63 billion dollars.

At the end of September Meta held 82.45 billion dollars in cash and equivalents. Long-term debt was 80.92 billion dollars. Business Recorder reports Meta's future spending commitments now total about 825 billion dollars, driven by data centres and AI infrastructure. The company raised its yearly capital expenditure range to between 135 and 145 billion dollars.

What the number says about the market

A 95 billion dollar equity raise does not fund a chat product. It funds buildings, power, cooling and silicon. Even the largest cash flows in software can no longer cover a data centre build at this scale. Meta is funding its growth with shares.

A chip maker buying its customer's stock is a signal of its own. Nvidia, AMD and Broadcom all need Meta to keep paying. A 5 billion dollar equity cheque alongside a hardware order is one way to say the demand will stay. The chip makers also share any gain if the shares rise.

Why this is a floor on inference prices

A buyer of AI inference should read this as a floor on prices. Hyperscale vendors are funding their compute with equity now, not cash flow. That cost must be earned back in the margin on tokens over the next decade. The companies renting those tokens will have to charge enough to pay for the data centres being built this week.

For a team deciding how to build an AI product, the message is simple. The market will fund the infrastructure for the long run. The compute will arrive. The pricing on it will reflect what each hyperscaler needs to earn. Meta, Microsoft, Amazon and Google all have shareholders to pay back.

Source

Business Recorder, as reported by Reuters, with additional detail from The Information.

This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.

Share
LinkedInX
Start a project