Leaked Meta Memo Reveals Plans to Double AI Computing Capacity to 14 Gigawatts – NaturalNews.com
Reuters reported an internal Meta memo outlining plans to double Meta’s AI computing capacity to 14 gigawatts by 2027, including 7 GW in 2027, plus long-term supply contracts for memory, flash storage, and fiber-optic equipment amid a shortage. The memo also describes in-house AI accelerator production with TSMC and design partner Broadcom, and an AMD accelerator supply agreement. Meta shares fell 4.3% at the open.
How this was made
The 30-second read
Why it matters
Traders can treat this as a capex-intensity and financing-risk catalyst: it reframes AI infrastructure spending as larger and more committed than investors may have priced, contributing to the immediate selloff and raising questions about FCF and leverage.
Market read
A leaked memo describing a major AI capacity expansion is linked to a same-day stock drop and to concerns about capex, off-balance-sheet commitments, and financing.
What to watch
The article emphasizes headline capex versus purchase obligations and leased compute, but does not quantify expected revenue ramp or margins from the incremental capacity, which could materially change the net valuation impact.
Background
The piece centers on a Reuters-reported leaked internal memo describing Meta’s next phase of AI infrastructure, including a 14 GW target by 2027 and a silicon supply cadence.
Ticker impact
Meta shares fell 4.3% after Reuters reported an internal memo to double AI computing capacity to 14 GW by 2027, including 7 GW this year.
Bearish bias for the next several sessions as investors reprice capex intensity versus capital-discipline expectations.
The article ties the stock’s open decline to the reported capacity expansion, adds cost/financing concerns (off-balance-sheet commitments, flat-to-negative 2026 FCF model), and frames the plan as contradicting prior market preference for discipline.
Market effects
Reinforces the hyperscaler AI capex boom narrative while highlighting memory and power-grid bottlenecks that can tighten supply and lift costs across the AI infrastructure supply chain.
US data-center power and water constraints could increase permitting and build-out friction, affecting AI infrastructure timelines.
Long-term component and fiber contracts underscore global supply-chain leverage (memory, storage, networking) and may shift bargaining power among hardware vendors.
Counterpoint
The capacity build could be a competitive necessity and may be offset by monetization via Meta Compute and improved utilization, limiting downside versus the market’s capex fear.
Key entities
- companyMeta Platforms Inc.
Subject of the leaked memo, with plans to double AI computing capacity to 14 GW by 2027 and detailed supply and chip strategy.
- mediaReuters
Reported the contents of the internal memo that drove the market reaction described in the article.
- personMark Zuckerberg
Referenced for prior town-hall remarks that agent development did not accelerate as expected, providing context for the stock’s sensitivity to capex plans.
