Meta's Reality Labs Has Lost $83.5 Billion Over 21 Quarters

When Meta released its quarterly earnings report on Wednesday evening, a colleague pointed out how Meta lost $4 billion on Reality Labs, the division responsible for its AR glasses, VR headsets, and VR software. I yawned at first. Meta losing $4 billion on Reality Labs just didn't seem surprising. It's a given. Reality Labs lost another $4 billion, and also, the sky is blue. Then I realized, that itself is notable — for Meta, losses on this unit are quite literally average behavior. Over its last 21 quarterly earnings reports, dating back to 2021, Meta has lost a total of $83.5 billion on Reality Labs, which comes out to an average of about $4 billion in losses each quarter. That is bananas! Equally astounding is that as Meta pulls back from its metaverse ambitions, its spending on AI will be even more astronomical. True, it's not like Meta doesn't have the money. In the first quarter of this year, the social media giant posted a net income of $26.8 billion, up 61% over the year prior; revenue also increased 33% year-over-year to $56.3 billion. But despite its foundation in social media, Meta's current goal is to stay competitive with AI leaders like OpenAI and Anthropic. Meta projected that it will spend between $125 billion and $145 billion in 2026, exceeding analysts' projections and Meta's previous estimates. "We are increasing our infrastructure capex forecast for this year," Meta CEO Mark Zuckerberg said on a public call with investors on Wednesday. "Most of that is due to higher component costs, particularly memory pricing [...] We are very focused on increasing the efficiency of our investments." Meta also spent a lot of money to build a metaverse that no one really wanted or cared about. It's going to take even more money to build an AI superintelligence that (maybe some) people actually want. Last year, Meta went on an expensive hiring spree, poaching over 50 AI researchers and engineers from competitors, which helped the company ship its newly overhauled AI model, Muse Spark, earlier this month. While CEO Mark Zuckerberg reported "large increases" in Meta AI use since that release, it's only getting more expensive to build and maintain AI products. On the earnings call, one concerned investor asked if Meta could provide an outlook for its 2027 capital expenditures. The response wasn't reassuring. "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," replied Meta CFO Susan Li. "Our experience so far has been that we have continued to underestimate our compute needs."