Meta $95 Billion AI Infrastructure Stock Offering, October 2026

Meta plans to sell up to $95 billion in new non-voting Class C shares — reportedly a record equity offering — to fund capital expenditure that reached $69.69 billion in FY2025 and is now guided as high as $145 billion in FY2026.

Meta's announcement this month that it plans to raise up to $95 billion by selling newly created non-voting Class C shares — a transaction Reuters described as a record equity offering — brings the company's AI infrastructure ambitions into sharp relief. The move, corroborated by outlets including CNBC, UPI, and the New York Times, follows an extraordinary escalation in capital spending: Slicast's own XBRL-derived compilation shows capital expenditure reached $69.69 billion in FY2025, an 87.1% increase from the $37.26 billion spent in FY2024, and equivalent to 35% of Meta's $200.97 billion in FY2025 revenue — the third-highest capex intensity among the seven hyperscalers Slicast tracks, behind only Oracle. An 8-K filed with the SEC on October 21 confirms that Meta's board approved, on October 19, a material definitive agreement on the unanimous recommendation of a special committee of independent directors, involving a proposed amendment and restatement, according to the filing summary. Multiple coverage items link that action to the Class C share creation, which is designed to preserve an uncapped fundraising channel without diluting existing voting control.

The timing is laden with context. Meta's Q3 2026 results showed revenue of $64.6 billion, up 26% year-over-year — a top-line beat — but multiple reports simultaneously flagged that quarterly capital expenditure came in at $34.9 billion, that full-year capex guidance was raised to as much as $145 billion, and that free cash flow fell sharply, with several outlets characterising the quarter as producing negative or near-zero free cash flow for the first time in at least a decade. Multiple reports also linked these earnings dynamics to a 7% decline in Meta's share price on the trading session of October 21, when the stock closed at $648.71. By the session of October 26, no specific cause was reported for a further retreat, with the shares settling at $630.56, down 1.9% on that day. The earnings ripple extended to infrastructure suppliers: GE Vernova fell 11%, Vertiv Holdings dropped 7%, and Eaton slid 6%, according to 24/7 Wall St., which cited Meta's spending outlook as stoking AI financing concerns.

The infrastructure being financed is already substantial and continues to expand. Meta's Prometheus campus in Ohio — its first phase now live — is reported to be actively used to train the Muse model, making it one operational node in a rapidly growing global footprint. The Hyperion campus in Richland Parish, Louisiana, has reportedly been scaled to a capacity of 5 gigawatts with a reported price tag above $50 billion, placing it among the largest data centre projects on record; Wall Street has separately arranged $27 billion in project financing for Hyperion. A 3,000-acre site in Mason County, Kentucky — publicly confirmed as Meta's — reportedly carries a 40-year property-tax abatement. Meta is also reported to be deploying compute capacity in temporary structures, a reflection of how urgently near-term demand outpaces permanent construction schedules. At the OCP Summit this month, Meta unveiled IcePack liquid-cooled racks and a 102.4-terabit IceCube switch, hardware whose specifications highlight the thermal and networking density demanded by large-scale AI training. Meta's infrastructure chief was separately cited as projecting the company's 2026 AI spending above $100 billion.

The competitive environment in which Meta is operating heightens the stakes considerably. Combined hyperscaler AI capital expenditure for 2026 has been tallied at $725 billion across Microsoft, Meta, Google, Amazon, Oracle, and peers, with one estimate placing Amazon's standalone target at $220 billion. Slicast's data rank Meta fourth in absolute capex among six tracked hyperscalers, and third in capex intensity among seven. According to Yahoo Finance and the New York Times, SpaceX is simultaneously pursuing a major fundraise for AI infrastructure alongside Meta, with the two reportedly seeking a combined $135 billion in new capital. Elon Musk, in comments cited by Yahoo Finance, warned of a potential funding crunch for neocloud operators — a caution that sits in direct tension with the scale of capital being mobilised by the largest incumbents. Slicast's records show Meta's highest single quarterly capital expenditure was $19.00 billion, for the quarter ended March 31, 2026.

The strategic logic of the buildout is not without friction. U.S. Congressional scrutiny has been directed at Meta's $6.6 billion in research tax credits claimed during the AI expansion, raising the prospect of legislative or regulatory pushback on the financing structure. Meta's reported departure from the RE100 renewable-energy pledge sits uneasily against its sustainability commitments, as gigawatt-scale electricity demand presses against both grid capacity and the available supply of clean power. The compute cost of operating frontier AI at scale is significant in its own right: tech-insider.org reported that Meta Muse would require $2.8 billion in compute resources to serve 100 million users. One analysis described Meta's cumulative infrastructure commitments as implying an AI compute cycle outlay approaching $825 billion, while noting that cost divergences across markets and funding gaps are widening — a picture of structural resource strain that extends well beyond any single company.

For analysts tracking Meta's trajectory, three signals will be determinative in the quarters ahead. First, the pricing and reception of the Class C offering: whether $95 billion can be raised at tolerable dilution will directly test equity-market appetite for AI infrastructure funding at this scale, particularly after the earnings-session sell-off. Second, free cash flow recovery: with quarterly capex at $34.9 billion running against Q3 revenue of $64.6 billion, AI-enhanced advertising, Muse, or other products will need to show a monetisation inflection before investors can accommodate what may be sustained cash consumption during a $145 billion build year. Third, the regulatory and energy environment: Congressional scrutiny of tax treatment, the power-procurement dynamics that follow Meta's post-RE100 stance, and Europe's cloud-sovereignty policy evolution — a lobby group backed by companies including Amazon, Google, and Meta warned this month that displacing American platforms could cost European economies €115 billion — each represent policy variables capable of materially reshaping the economics of what is now one of the largest capital programmes in corporate history. The ambition is, at this point, precisely quantifiable; whether the returns will match it remains an open question.