M. V. Cunha
Equity Analyst, Long-Term Investor, Fundamental Valuation
Meta's compute is not excess capacity; it needs all it can get, so the selloff on that headline was noise. Reuters reported Meta plans to double its computing capacity to 14 GW by 2027, and Mark Zuckerberg said Meta needs all the computing power it can get. In a compute-starved market, external offers are so high that Meta may sometimes rent out capacity instead, and capacity resold above its cost is not excess. x.com x.comÂ
Meta's $35B financing plan points to continued chip demand rather than a capex cut. Meta is raising money to buy more chips because it is short of them, contrary to earlier claims that it lacked demand for its chips. x.comÂ
Meta's unmet AI demand pays its bottleneck suppliers. Super Micro's $19.5B quarterly outlook and under 8x forward earnings point to a double. CoreWeave's Meta deals total $35.2B; it should easily top $100. Builders say nearly all new sites use Dell or Super Micro; Dell could easily double. Micron's demand is off the charts at 5x forward earnings. Sources say Vertiv's gear is sold out for 2-3 years. VivoPower has 2GW+ potential in 5-7 years. Digi Power X has a $1.1B Cerebras deal. x.com x.com … Show more
Meta's swelling construction in progress has not entered the depreciation base yet. It reached $89.5B in September, up from $26.8B at the end of 2024, largely from data centers, servers and network infrastructure. At its recurring quarterly pace, it is highly likely, by this estimate, to surpass $100B by year-end, nearly quadrupling in two years, a roughly $73B increase. x.comÂ
Meta's plan to build a new $10B data center in Canada to expand AI capacity undercuts the media framing that its Meta Compute effort was overbuilding and would cut capex. x.comÂ
Meta's voluntary move is bullish because it expects to profit from it. Its massive debt capacity remains untouched and is increasing. x.comÂ
Large-scale shareholder financing makes Meta's AI buildout more concerning after quarterly free cash flow turned negative. The WSJ reports that Meta is looking to raise $95 billion: $35 billion in common stock and convertible preferred now, followed by up to $60 billion from 2027. x.comÂ
Meta's negative free cash flow from AI capex is an expected investment phase, not a broken thesis. Analysts had forecast deterioration through 2027 as infrastructure spending accelerates, followed by a sharp recovery in 2028 as those investments begin generating cash. The market is treating known spending as a surprise, creating an opportunity while Meta builds infrastructure for the next decade of AI. x.comÂ
Meta's financial strength gives it an advantage in securing scarce capital for AI infrastructure, with funding going to those trusted to earn the best returns. Meta says its planned $60B at-the-market equity program is not expected to begin until Q1 2027 and complements $35B in underwritten offerings; its Aa3/AA- ratings allow efficient long-term borrowing. x.comÂ
Meta's hundreds-of-gigawatts compute ambition looks hard to justify against the revenue a much smaller build might support. If Anthropic can generate over $70B in annualized revenue on 3GW or less, including training, hundreds of gigawatts is not a serious figure. Earlier Muse monetization than in past product cycles also makes little sense without direct integration into ad recommendations or an immediate shift in user behavior. x.comÂ
Meta can monetize the same compute twice, which makes owned, powered compute its moat. Rothschild expects Meta's power use to grow from 2.5 GW to 15 GW by 2030, much of it excess because it thinks Meta will launch a cloud business. AI first pays for each gigawatt internally through a stickier feed and better ad targeting; leftover compute can then be sold as an AI cloud at cloud margins. In a power-constrained world anyone can rent GPUs, but almost nobody can energize new gigawatts on demand. … Show more