Meta AI spending soars to $145 billion as Zuckerberg faces compute dilemma and investor revolt

Reviewed byNidhi Govil

19 Sources

Share

Meta raised its AI spending forecast to between $130 billion and $145 billion for 2026, but free cash flow collapsed 91% to just $784 million. Mark Zuckerberg now faces a critical choice: rent out scarce AI compute capacity for immediate returns or preserve it for Meta's own AI ambitions. Investors are growing skeptical as the company spends like a cloud giant without a cloud business model.

Meta AI Spending Surges as Free Cash Flow Collapses

Mark Zuckerberg is doubling down on Meta AI spending even as investors punish the company for its financial strain. Meta raised the low end of its capital expenditure guidance by $5 billion, bringing the 2026 range to between $130 billion and $145 billion, mostly directed toward AI infrastructure including chips, servers, energy, and data centers

1

. The company's free cash flow plummeted 91% to just $784 million in the second quarter, down from $8.55 billion a year earlier

4

. Meta's stock dropped 11% following the earnings announcement, reflecting growing investor skepticism about the company's ability to generate returns on its massive AI investment

3

.

Source: New York Post

Source: New York Post

The spending spree comes as Meta reported second-quarter revenue of $60.8 billion, up 28% from a year earlier, its fastest revenue growth since late 2021

5

. Despite strong top-line performance driven by AI-tuned ad recommendations that kept users scrolling on Instagram and Facebook, earnings of $6.18 a share missed analyst expectations of $7.22

5

.

The AI Capacity Dilemma: Sell or Keep?

Mark Zuckerberg revealed during the earnings call that Meta faces a critical AI capacity dilemma. The company has received numerous offers for its AI compute at "a significant premium" over what it paid to build that capacity

2

. However, renting out compute could divert scarce resources from Meta's own push to build AI models and enterprise AI services.

"A common trade-off that we need to make is around how much do you monetize something today versus develop future assets," Zuckerberg explained on the call

2

. He framed AI compute as a scarce strategic asset that the company should keep and build around, rather than simply sell for short-term profit. "It would be foolish to basically just sell all of the compute and take a short-term profit," he added

2

.

As CNBC reported earlier this month, Anthropic is in preliminary talks to lease computing power from Meta

2

, signaling that the company is actively exploring opportunities to monetize its infrastructure.

Spending Like a Hyperscaler Without the Business Model

Meta's challenge is unique among Big Tech companies. While it's spending at levels comparable to hyperscalers like Microsoft, Alphabet, and Amazon, it lacks their cloud infrastructure business to immediately monetize that investment. "Meta is spending like a hyperscaler without a hyperscaler's business model," said Josh Gilbert, eToro's lead APAC analyst

1

. "Microsoft, Alphabet and Amazon can point their data centre dollars at cloud businesses that sell compute straight back out the door, but Meta doesn't have the same outlet, so every dollar of build-out leans on the ads business."

Source: Reuters

Source: Reuters

Meta CFO Susan Li defended the AI spending plans, saying the industry had "underbuilt historically" for AI demand, making existing capacity "extremely valuable"

1

. She said capacity would remain tight "for the foreseeable future," creating opportunities for Meta to generate returns through products, enterprise services, and compute sales.

Zuckerberg argued that "there will continue to be a significantly higher margin on selling intelligence rather than selling compute directly, but we think that there's a big opportunity obviously to sell compute as well"

1

.

AI Agents and Enterprise Push Signal New Direction

Mark Zuckerberg outlined Meta's AI spending plans, emphasizing that AI agents will be "the next wave of our product line in the months and years to come"

3

. "Soon, we'll have agents that can work 24/7 on your behalf," he said during the call, promising that great personal AI agents need to "just work out of the box" for billions of people to adopt

3

.

Source: ET

Source: ET

Earlier this month, Meta debuted the Muse Spark 1.1 model under the leadership of AI chief Alexandr Wang, who called it the "strongest model for agentic and coding work yet" at a cheaper price than offerings from OpenAI and Anthropic

2

.

Zuckerberg acknowledged that building an enterprise business will require "a different muscle than we've historically had"

3

. The company recently confirmed that Dave Brown, a former longtime senior executive at Amazon Web Services, is set to join Meta

2

, signaling serious intent to build enterprise capabilities.

Meta is already using LLM technology to improve ad recommendations and algorithms across its social media platforms. "Earlier this year, we reached a milestone of every public Reels and Feed post on Instagram being automatically processed through an LLM and analyzed across dimensions from topic to tone," CFO Susan Li said

4

.

Metaverse Echoes and Legal Troubles Compound Concerns

Investor skepticism is amplified by memories of Zuckerberg's metaverse bet, which continues to drain resources. Meta's Reality Labs unit lost $4.62 billion in the latest quarter on just $431 million of revenue

2

, and has generated more than $80 billion in total operating losses over roughly six years

4

.

Adding to Meta's challenges, the company spent $2.4 billion in charges related to legal proceedings in the past quarter

4

. The bulk of lawsuits involve accusations that Meta's platforms were designed to addict young users, with four states—California, New Jersey, Colorado, and Kentucky—seeking up to $1.4 trillion in damages

4

.

"I get that this is sort of a big bet across the industry," Zuckerberg said. "My personal bet is that the people who invest in this are going to be rewarded and feel very good over time"

4

. Whether investors share that confidence remains an open question as Meta's long-term bet on AI continues to test their patience.

Today's Top Stories

© 2026 TheOutpost.AI All rights reserved