Meta's $145 Billion AI Spending Under Scrutiny as Q3 Earnings Loom

6 Sources

Share

Meta reports Q3 earnings Wednesday with investors demanding proof that massive AI investments will pay off. The company's Muse AI assistant gained 2.8 million downloads in 12 days, but capital expenditures reached $145 billion in 2026 while free cash flow dropped 91% to $784 million. Wall Street questions whether Mark Zuckerberg's bold AI infrastructure buildout can justify mounting costs.

Meta Faces Critical Test on AI Investment Returns

Meta reports third-quarter earnings Wednesday after the bell, with analysts expecting earnings of $5.75 per share on revenue of $63.2 billion

1

. The earnings call comes as investors grow increasingly anxious about whether Mark Zuckerberg's aggressive AI spending will translate into sustainable revenue growth. Meta narrowed its 2026 capital expenditures guidance to between $130 billion and $145 billion in July, representing a $5 billion increase to the low end of its prior range

1

. This massive AI infrastructure buildout resulted in free cash flow shrinking to $784 million from $8.55 billion in the same quarter a year earlier

1

.

The company's stock has suffered a five-session losing streak and closed in the red in nine of the last 11 trading days, positioning Meta for its worst annual performance since 2022

2

. Wall Street anticipates Meta will post capital expenditures of more than $40 billion in the third quarter alone, compared with less than $19 billion a year ago

2

. Analysts project capex will reach $196.3 billion in 2027 and $218 billion in 2028

1

2

.

Muse AI Agent Shows Early Promise Amid Market Skepticism

Meta's Muse AI personal assistant tool debuted in September and shot to the top of Apple's iOS app store, helping Meta end September with its best month on Wall Street since 2022

1

. The Muse AI assistant attracted 2.8 million downloads in its first 12 days and gained 2.5 million daily users

3

. Citi analysts believe the app could eventually add at least $27 billion in annual revenue by fiscal 2030

4

.

Meta's AI strategy centers on its foundation models, with AI chief Alexandr Wang continuing to roll out updates at a rapid pace. The company released Muse Wonder and Muse Wonder Full on October 12, marking its first serious frontier model since Wang reorganized Meta's AI work in August 2025

4

. Both models were trained at Meta's Prometheus data center in central Ohio, with the facility now operational and plans to expand toward one million Nvidia GPUs

3

. Wang revealed that building a frontier model now costs tens of billions of dollars

4

.

Wall Street Questions Sustainability of AI Trade

Andrew Wang, managing partner at Runnymede Capital Management, stated that a successful report from Meta would improve confidence in the market and enliven the AI trade by easing concerns about demand across the industry

2

. However, a concerning report would put more of a cloud over the sector. The Philadelphia Stock Exchange Semiconductor Index has lost 21% in the second half of the year as investors question the durability of AI capex programs

2

.

Meta's spending has direct implications for its supply chain partners. Advanced Micro Devices relies on Meta for 5.5% of its revenue, making Meta AMD's second-largest customer behind HP

2

. Muse's success has given Wall Street analysts confidence in AMD's outlook, demonstrating how sentiment toward chipmakers hinges on big AI spenders remaining committed to their plans. Meta is also developing four generations of custom AI chips with Broadcom to reduce computing costs and reliance on Nvidia

3

.

Massive Data Center Expansion Raises Financial Concerns

Meta's Hyperion data center in Richland Parish, Louisiana, will grow to 5 gigawatts of computing capacity from an original design of 2 gigawatts, pushing the cost above $50 billion from a prior estimate of $27 billion

4

. The facility will be Meta's largest data center. Meta also revealed a five-year computing agreement with Nebius Group worth up to $27 billion, including $12 billion in dedicated capacity beginning in 2027, plus up to $15 billion in additional purchases

3

.

Source: Seattle Times

Source: Seattle Times

Speculation is growing that Meta will issue new stock to finance its massive AI infrastructure buildout

5

. Short interest positioning in Meta is up over 20% or 33 million shares since mid-August, with traders betting on a potential dilutive secondary stock sale

5

. Meta has $90 billion in cash but faces $145 billion in spending this year alone, leading some analysts to anticipate a major capital raise

5

.

Strategic Shifts and Revenue Diversification

Mark Zuckerberg outlined a path beyond simply renting out server capacity during Meta's second-quarter earnings call, stating there will continue to be a significantly higher margin on selling intelligence rather than selling compute directly

4

. Meta disclosed that its new Generative Ads Model improved conversions by 18.9% and click-through rates by 12%

3

. Ad revenue generated $59.4 billion of Meta's $60.8 billion in second-quarter revenue

3

.

Source: Benzinga

Source: Benzinga

Meta's relationship with Apple has also evolved. On October 15, Zuckerberg publicly thanked Apple CEO John Ternus for helping Meta's Muse Spark Open model run on Mac Studio computers, marking the first public partnership between the two companies in years

4

. This represents a significant shift from February 2022, when Meta estimated Apple's App Tracking Transparency changes cost the company $10 billion

4

. Bank of America analyst Wamsi Mohan warned that AI agents could capture shopping, referrals and payments that now flow through Apple's ecosystem, potentially shifting the power balance

4

.

Today's Top Stories

© 2026 TheOutpost.AI All rights reserved