The Lombard Review हिन्दी

Meta spends more, investors flee

DOG-52 diodes and transistors made by Tewa in an Odra 1002 computer.
DOG-52 diodes and transistors made by Tewa in an Odra 1002 computer. Photo: Topory/Wikimedia Commons · CC BY-SA 3.0

Meta Platforms shares fell 10.5 per cent in a single session after first-quarter results that were, on the face of it, solid. Revenue and earnings were fine. What spooked investors was Meta raising its full-year capital expenditure guidance to $35–40 billion without lifting its near-term revenue forecast to match. Wall Street's patience with Mark Zuckerberg's spending has limits.

From efficiency to ambition

The reaction makes more sense against the past two years. In 2022 the shares lost about two-thirds of their value as advertising slowed, Apple's privacy changes hurt targeting and Reality Labs, the metaverse division, ran up losses measured in the tens of billions. Zuckerberg responded by declaring 2023 a "year of efficiency", cutting some 21,000 jobs across two rounds of lay-offs and trimming spending plans. Investors rewarded the discipline: the stock nearly tripled last year.

The rally ran further this year. In February Meta announced its first dividend and a $50 billion buyback alongside strong fourth-quarter results, and the shares jumped by about a fifth in a day. That kind of move sets expectations, and it attracted a shareholder base that bought a cash machine run with restraint. The first-quarter release told those shareholders that the restraint was over.

Spending before selling

Investors who applauded Meta's 'year of efficiency' did not enjoy the return of big capital commitments, this time for AI infrastructure with no clear payback date. Zuckerberg asked shareholders to look past the near-term spending to eventual monetization. They remember the metaverse, which burned through billions of dollars with little to show for it. Heavy spending without visible revenue usually costs a stock its multiple.

The spending is real and front-loaded. Zuckerberg said in January that Meta expected to own about 350,000 of Nvidia's H100 accelerators by the end of this year, and the company released its Llama 3 models only a week before the results. It also warned that capital expenditure would keep rising next year. Data centres and chips are depreciated over several years, so today's outlays will weigh on reported margins well after the cash has gone out of the door.

The rest of the guidance compounded the problem. Meta put second-quarter revenue at between $36.5 billion and $39 billion, a range whose midpoint sat below analysts' expectations, and nudged up the bottom of its forecast for full-year expenses. Reality Labs lost about $3.8 billion in the first quarter alone. Put together, the message was higher costs arriving now and extra revenue arriving at some unspecified later date.

How the market does the sums

The arithmetic behind the sell-off is simple. Investors increasingly value mature technology companies on free cash flow, and capital expenditure comes straight out of it. Every extra billion spent on servers is a billion not available for buybacks or dividends this year, and the depreciation that follows depresses reported earnings for years afterwards. A company that raises spending without raising its revenue outlook is, in effect, telling the market that its cash yield will fall before it rises.

The Colonial National Bank Building, Connellsville, Pennsylvania.
The Colonial National Bank Building, Connellsville, Pennsylvania. Photo: Canadian2006/Wikimedia Commons · CC BY-SA 3.0

Why rivals were forgiven

The contrast with Microsoft and Alphabet, which reported a day later, is instructive. Both are also spending heavily on AI hardware, yet their shares rose, and Alphabet's jumped after it announced its own first dividend. The difference is that both sell computing capacity to other companies. Their cloud divisions can point to AI demand showing up in revenue within the same quarter as the spending.

Meta has no such shop window. It rents no computing to outsiders, gives its Llama models away and makes almost all its money from advertising. The payback from AI therefore has to come indirectly, through better ad targeting, more engaging feeds and, eventually, products that users or businesses pay for. Each of those may prove lucrative, but none can be measured cleanly from the outside, and investors discount what they cannot measure.

The case for Zuckerberg

The strongest defence is that Meta can afford the bet. The core advertising business grew revenue by more than a quarter in the first quarter and throws off enough free cash to fund the capital programme, the dividend and the buyback at once. AI-driven recommendations have already helped Reels and the automated ad tools that many advertisers now use. On this reading, the market is punishing Meta for being early and honest about costs that its rivals will also incur.

That argument has force, and it may well be vindicated over a period of years. It also rests on a long horizon at a time when investors are being paid more than five per cent to hold cash, and it asks shareholders to trust the same management that promised a metaverse payoff. The October 2022 sell-off, when Meta lost about a quarter of its value in a day on a similar mix of rising costs and soft guidance, shows how quickly that trust can be withdrawn.

Who gains from the outlay

The clearest winners from Meta's plans sit on the other side of the invoice. Nvidia and the makers of servers, networking gear and memory are the direct beneficiaries, and electricity suppliers and data-centre builders gain from the demand for power and space. Advertisers may benefit too if better targeting lowers their costs per sale. The shareholders paying for it all are the only group being asked to wait.

The episode matters beyond Meta. Microsoft, Alphabet, Amazon and Meta between them place a large share of the orders sustaining the AI hardware boom, and their investors are now signalling that they will fund the spending only where revenue follows. If that discipline spreads, the suppliers' order books will be the first place it shows.

What would change the view

The next evidence arrives with second-quarter results in late July. Investors will look for revenue growth holding up against tougher comparisons, any sign that Meta's AI assistant or business messaging tools are producing income, and whether the capital guidance for next year is framed with a revenue story attached. A slowdown in Reality Labs losses would also help rebuild credibility.

Meta's shareholders want AI revenue they can see within a few quarters. Until Zuckerberg can show the spending paying off, the shares will probably stay under pressure, and the July results are the earliest point at which he can make that case with numbers rather than promises.

Write to Arnav Goel at contact@thelombardreview.com

More From The Lombard Review