Inside Meta's $145 Billion AI Infrastructure Bet and the Cloud Ambitions Zuckerberg Won't Quite Confirm
Meta has raised its 2026 capital expenditure guidance to between $125 billion and $145 billion, up from $72.2 billion in 2025, as it builds AI infrastructure at a scale rivaling AWS, Azure, and Google Cloud. At the Q1 2026 earnings call on April 29, Zuckerberg called questions about AI ROI "very technical" while CFO...
What did Mark Zuckerberg say at Meta's May 2026 shareholder meeting about potentially entering the cloud computing market, what is the curreMeta is building AI infrastructure that rivals the largest cloud providers, with 2026 capex now reaching up to $145 billion.
Prompt de IA
Create a landscape editorial hero image for this Studio Global article: What did Mark Zuckerberg say at Meta's May 2026 shareholder meeting about potentially entering the cloud computing market, what is the curre. Article summary: ## What Zuckerberg Said at the May 2026 Shareholder Meeting. Topic tags: general, general web, user generated, government. Reference image context from search candidates: Reference image 1: visual subject "On Wednesday's earnings call, Meta CEO Mark Zuckerberg outlined several of the ways the company plans to capitalize on this year's "major AI" source context "Meta to 'push the frontier' in 2026: CEO Zuckerberg explains how" Reference image 2: visual subject "Mark Zuckerberg, chief executive officer of Meta Platforms Inc., during the Meta Connect event in Menlo Park, California, US, on Wednesday, Sept. 25, 2024. Meta Platforms Inc. debu" source context "Mark Z
openai.com
Meta Platforms is in the middle of the most aggressive AI infrastructure buildout in corporate history. The company's capital expenditure guidance now reaches as high as $145 billion for 2026 — a figure that puts it squarely in the same league as Amazon Web Services, Microsoft Azure, and Google Cloud. That has sparked a persistent question across Silicon Valley and Wall Street: is Meta quietly positioning itself to become the fourth major cloud provider?
Zuckerberg hasn't announced a "Meta Cloud Console" yet. But in earnings calls, shareholder meetings, and strategic moves over the past year, the signals have grown louder. Here's what was actually said — and what the spending numbers reveal.
What Zuckerberg Has Actually Said About Selling Compute
The idea that Meta might sell access to its AI infrastructure didn't start at the May 2026 annual shareholder meeting, though that gathering — held virtually on May 27, 2026 — is the most recent venue where investors pressed for clarity. Detailed public reporting on Zuckerberg's specific comments about entering the cloud market at that meeting is still limited as of late May 2026.
But the trail of public statements leading up to it is unmistakable:
October 2025, Q3 2025 earnings call: Asked whether Meta could sell its excess compute capacity, Zuckerberg replied, "People keep asking if we can share compute — we haven't done that yet, but that is a reasonable thing for people to be asking about". Analysts and industry observers widely interpreted this as the strongest hint yet that selling AI compute is on the table.
April 29, 2026, Q1 2026 earnings call: When an analyst pressed for signs of return on investment from the ballooning infrastructure spend, Zuckerberg deflected with "that's a very technical question". The company instead pointed to long-term AI product cycles and agent-driven business transformation.
Studio Global AI
Search, cite, and publish your own answer
Use this topic as a starting point for a fresh source-backed answer, then compare citations before you share it.
What is the short answer to "Inside Meta's $145 Billion AI Infrastructure Bet and the Cloud Ambitions Zuckerberg Won't Quite Confirm"?
Meta has raised its 2026 capital expenditure guidance to between $125 billion and $145 billion, up from $72.2 billion in 2025, as it builds AI infrastructure at a scale rivaling AWS, Azure, and Google Cloud.
What are the key points to validate first?
Meta has raised its 2026 capital expenditure guidance to between $125 billion and $145 billion, up from $72.2 billion in 2025, as it builds AI infrastructure at a scale rivaling AWS, Azure, and Google Cloud. At the Q1 2026 earnings call on April 29, Zuckerberg called questions about AI ROI "very technical" while CFO Susan Li cited rising component costs and data center expansion as the main drivers behind the latest $10 b...
What should I do next in practice?
Meta is simultaneously a major cloud customer and potential competitor: it signed a $10 billion, six year deal with Google Cloud in August 2025 for AI workloads, while maintaining its legacy "long term strategic" rela...
May 2025 annual shareholder meeting: A year earlier, Zuckerberg took a different angle on AI monetization. He revealed that Meta AI had reached 1 billion monthly active users and said the company may explore subscriptions or paid AI tiers that offer a subscription service granting access to more compute .
The arc of these comments matters. In 2025, monetization meant paid AI assistants. By late 2025 and into 2026, the language shifted toward selling raw compute — effectively, cloud infrastructure. No formal product has been announced, but the door is now openly acknowledged.
Meta's 2026 AI Infrastructure Spending: $145 Billion and Climbing
Meta's capex trajectory is the clearest signal of its infrastructure ambitions. The company has raised its guidance repeatedly, with the latest revision coming during the Q1 2026 earnings report on April 29, 2026.
2026 full-year capex (latest guidance): $125 billion to $145 billion
Prior guidance (early 2026): $115 billion to $135 billion
2025 actual spend: $72.2 billion
2024 actual spend: $39 billion (approximately)
Cumulative US infrastructure plan by 2028: ~$600 billion
CFO Susan Li attributed the latest $10 billion revision to "anticipated increases in component costs" and "additional data center costs to support future-year capacity". Total expenses for 2026 are forecast at $162 billion to $169 billion, meaning capex now represents roughly three-quarters of Meta's total spending .
At the midpoint of $135 billion, Meta's 2026 capital spending would exceed the GDP of over 120 countries . The company is building gigawatt-scale AI clusters, deploying its own custom silicon called MTIA (Meta Training and Inference Accelerator), and racing to stand up new data center campuses worldwide. In the words of one analysis, Meta is constructing infrastructure that rivals the biggest clouds on earth — without offering a cloud service yet .
Which Cloud Providers Does Meta Currently Use?
Meta's cloud relationships are a study in duality: the company is simultaneously a top-tier customer of the existing hyperscalers and a potential future competitor.
Google Cloud — The $10 Billion AI Partnership
In August 2025, Meta signed a landmark $10 billion, six-year deal with Google Cloud, first reported by The Information and later confirmed by multiple outlets . The deal gives Meta access to Google's custom TPU chips and the Vertex AI platform to accelerate development and scaling of Meta's Llama models.
This strategic shift meaningfully reduced Meta's historical reliance on AWS and Azure for AI compute. It was also a significant win for Google Cloud, which has been gaining ground on its larger rivals in the AI era. Google Cloud revenue grew 32% year-over-year in Q2 2025, and the Meta deal is considered a defining moment in the intensifying cloud wars .
AWS — A "Key" Long-Term Strategic Provider
Meta designated AWS as a "key" and "long-term" strategic cloud provider in a formal announcement in late 2021 . That relationship persists, though the Google Cloud deal and Meta's accelerated buildout of its own data centers have shifted the balance. AWS remains a significant partner, but Meta is no longer as dependent on any single external cloud provider as it once was.
Microsoft Azure — A Less Documented Role
Meta's public reliance on Azure has been less pronounced compared to its relationships with Google Cloud and AWS. While Microsoft's enterprise reach and OpenAI integration make Azure an important player in the broader cloud ecosystem, Meta's infrastructure strategy has not leaned heavily on Azure for AI workloads, especially after cementing the Google deal.
The Bigger Picture: Customer, Competitor, or Both?
Meta is in an unusual position in the hyperscaler landscape. It is building AI infrastructure at a scale that rivals the Big Three cloud providers, yet it cannot log into a Meta Cloud Console to spin up a virtual machine today .
The company is a major customer — spending billions with Google and maintaining its AWS relationship — while simultaneously constructing the kind of capacity that could make it a seller. Its custom MTIA chips, gigawatt-scale data centers, and massive GPU clusters are the same building blocks AWS, Azure, and Google Cloud use to serve external customers.
Zuckerberg's public remarks suggest Meta is still weighing the option. On the Q3 2025 call, he said the company hasn't made a decision, but acknowledged selling compute as a "reasonable thing for people to be asking about". With capex now at $145 billion, the economic incentive to monetize unused or excess compute is only growing.
For now, Meta's infrastructure is an internal weapon — powering ad targeting, AI assistants, Llama model training, and agent-based experiences across its apps. But the foundations are being laid for something much larger. Whether Meta formally enters the cloud market or not, its infrastructure spending has already reshaped the competitive dynamics among the Big Three.
Comments
0 comments