Meta lost another $4 billion on Reality Labs in the first quarter of 2026, bringing total losses on its AR and VR division to $83.5 billion across 21 quarterly reports dating back to 2021. The company disclosed the figures Wednesday alongside Q1 net income of $26.8 billion, up 61% year-over-year, and revenue of $56.3 billion, up 33%. Investors were unimpressed: the stock fell more than 5% in after-hours trading.
The quarterly Reality Labs loss is now so consistent it functions as a constant. An average of $4 billion in losses per quarter, every quarter, for more than five years. The metaverse bet that Mark Zuckerberg renamed the company around in 2021 has produced no commercial breakout, and yet the spend rolls on.
What spooked the market was not the headset losses but the AI capex line. Meta now projects between $125 billion and $145 billion in infrastructure spending in 2026, above analyst expectations and above Meta's own prior guidance.
Key facts
- 01Meta lost $4B on Reality Labs in Q1 2026, bringing cumulative losses to $83.5B since 2021.
- 02Q1 2026 net income hit $26.8B, up 61% year-over-year, on revenue of $56.3B, up 33%.
- 03Meta raised its 2026 infrastructure capex forecast to $125B–$145B, exceeding prior guidance.
- 04Meta poached 50+ AI researchers and engineers last year to ship its Muse Spark model in April 2026.
- 05The stock fell more than 5% in after-hours trading despite the earnings beat.
"We are increasing our infrastructure capex forecast for this year," Zuckerberg told investors on the call. "Most of that is due to higher component costs, particularly memory pricing. We are very focused on increasing the efficiency of our investments." Memory pricing has tightened across the industry as hyperscalers compete for HBM supply.
“Reality Labs has lost $83.5 billion across 21 quarterly reports since 2021, an average of $4 billion per quarter, with no sign the bleed is slowing.”— Jaeden Schafer
The pivot from metaverse spending to AI spending is not a pivot in scale. It is an escalation. Reality Labs cost roughly $4 billion a quarter; AI infrastructure now costs that much in days. And unlike Reality Labs, the AI buildout is competing directly with OpenAI, Anthropic, and Google, all of whom are spending similar or larger sums.
Meta poached more than 50 AI researchers and engineers from competitors last year, an expensive hiring spree that helped produce Muse Spark, the overhauled Meta AI model that shipped earlier this month. Zuckerberg said usage of Meta AI has seen "large increases" since the release, though the company did not break out specific numbers.
An investor on the call pressed CFO Susan Li for a 2027 capex outlook. Her response did not soothe nerves. "We aren't providing a specific outlook for 2027 capex, and we are, frankly, undergoing a very dynamic planning process ourselves as we're working through what our capacity needs will be over the coming years," Li said. "Our experience so far has been that we have continued to underestimate our compute needs."
That admission — that Meta keeps underestimating compute demand — is the line that matters. It implies 2027 capex could land higher still, on top of a 2026 number that already approaches the GDP of a small country. The core advertising business is funding all of it, and for now the ad business is healthy enough to absorb the burn.
The skeptic's case is straightforward. Meta spent more than $80 billion on a metaverse that produced Quest sales and not much else, and the same management team is now writing checks an order of magnitude larger for AI superintelligence. Li's framing — that compute needs keep coming in higher than planned — is exactly what an investor worried about capital discipline does not want to hear.
The bull case is that AI, unlike VR, is already inside products billions of people use. Meta AI runs across Instagram, WhatsApp, and Facebook, and Muse Spark gives the company a frontier-class model to ship into that distribution. If even a fraction of that usage converts to ad lift or new revenue surfaces, the math works in a way the metaverse math never did.
Meta's quarter captures the strange shape of the current AI economy. A company can post a 61% jump in net income, raise its capex guide by tens of billions, lose $4 billion on a side project nobody asks about anymore, and still get punished by the market — because the only number investors are pricing now is how much more compute the next year will demand. On that question, even Meta's CFO is saying out loud that nobody knows.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
