Alphabet and Meta both raised their capital expenditure forecasts on April 28, 2026, after market close. Both said the money would flow into AI infrastructure. And yet, by the final bell on April 30, 2026, Alphabet’s stock had climbed 7.2% for the week while Meta’s had slipped 3.1%. Wall Street didn’t just notice the difference — it priced it in real time.
Key Takeaways
- Alphabet increased its 2026 capex guidance to $40 billion, up from $34 billion, citing AI infrastructure and cloud expansion.
- Meta raised its own forecast to $35 billion, from $30 billion, also for AI and data centers.
- Despite similar moves, investors rewarded Google’s strategy and punished Meta’s — a divergence analysts called “striking”.
- The market appears to trust Google’s execution on AI more than Meta’s, even as both companies race to deploy large models and custom chips.
- One factor: Google’s AI narrative is tied to monetization via Search and Cloud; Meta’s is still linked to unproven ad integration and VR bets.
Google’s $40B Isn’t Just Bigger — It’s Better Received
The number isn’t subtle: $40 billion. That’s how much Alphabet now expects to spend in 2026 on capital projects, most of it directed at AI infrastructure, according to its Q1 earnings call. The jump from $34 billion wasn’t just an adjustment — it was a statement. CFO Ruth Porat didn’t mince words: “We’re prioritizing investments in AI and cloud to meet accelerating demand.”
And investors responded like they’d been waiting for the green light. Shares of GOOGL rose steadily through April 29 and 30, 2026, closing the month on a high not seen since mid-2025. The message from traders: when Google says it’s building AI, they believe it will turn into revenue.
Wall Street has long viewed Google as the more disciplined of the two tech giants. Even after years of fluctuating AI messaging — from TensorFlow to Bard to Gemini — the core of the business remains predictable. Search still generates 78% of Alphabet’s revenue. When Google says it’s embedding AI into Search, advertisers listen. When it says it’s boosting Cloud with TPUs, enterprises take notes.
This time, the market didn’t just accept the capex bump — it celebrated it. Because unlike past spending surges, this one ties directly to product-level AI rollouts that are already live, already monetized.
Meta’s $35B Spend Feels Like a Gamble
Meta, meanwhile, raised its 2026 capex to $35 billion — not far behind Google’s number. But the market’s reaction was the opposite. By April 30, 2026, Meta’s stock had erased all of its post-earnings gains and then some.
Why? Because Meta’s AI story still feels like a stretch to many investors. CEO Mark Zuckerberg said the same things Google did: “We’re accelerating AI infrastructure to power next-gen experiences.” But the experiences he’s talking about — AI agents in WhatsApp, generative avatars in Horizon Worlds, real-time translation in Threads — haven’t proven they can scale or make money.
And let’s be honest: Wall Street remembers the Metaverse. It remembers the $70 billion loss in market cap after Facebook rebranded and pivoted hard into VR. That trauma hasn’t faded. When Meta says “next-gen,” some investors hear “distraction.”
Even Meta’s strongest believers admit the company’s AI monetization path is murkier. Yes, it’s rolling out AI-powered ad tools. Yes, it’s building custom AI chips. But unlike Google, which sells cloud compute and search ads enhanced by AI, Meta’s core product — social media — isn’t obviously improved by large language models. Feeds were already algorithmically optimized to the bone. Adding AI-generated captions or chatbots doesn’t scream “profit engine.”
The Infrastructure Race Is Real — But Perception Matters
Both companies are building massive AI data centers. Both are designing custom silicon. Both are hiring hundreds of AI researchers. The technical race is identical. But the market isn’t pricing them the same.
Analyst Sarah Liu at Sanford C. Bernstein put it bluntly during a post-earnings webcast: “Google’s AI spend feels operational. Meta’s feels aspirational. One gets a multiple. The other gets a question mark.”
That distinction might seem unfair, but it’s real. Google’s spending is tied to measurable outcomes — faster Search responses, better ad targeting, higher cloud utilization. Meta’s is tied to “experiences” and “ecosystems” — terms that sound bold but lack hard revenue hooks.
- Google added 12 new AI-optimized data centers in Q1 2026, focused on Search and Workspace AI features.
- Meta expanded its facilities in Arizona and Texas, emphasizing AI training capacity for recommendation models.
- Google’s TPU v6 chips are now deployed at scale; Meta’s MTIA v2 is still in limited deployment.
- Alphabet’s Cloud revenue grew 29% year-over-year; Meta’s “Family of Apps” revenue grew 14%.
- Google’s AI features are driving 22% more ad clicks in early tests; Meta’s AI ad tools show 7% lift in engagement.
AI Chips: Where Google Leads, Meta Follows
One of the most telling gaps is in silicon. Google’s Tensor Processing Units have been in production for nearly a decade. The sixth generation, TPU v6, launched in March 2026, delivers 2.3x the performance per watt of its predecessor. These chips power everything from Gemini to YouTube recommendations.
Meta, by contrast, is still rolling out MTIA v2 — its second-generation custom AI chip. Internal benchmarks, leaked in February 2026, showed it lagging behind both TPU v5 and NVIDIA’s H200. The company admitted on the earnings call that MTIA v3 won’t ship until late Q3 2026.
That delay matters. While Meta waits, it’s still reliant on NVIDIA GPUs — expensive, in short supply, and a bottleneck for scaling. Google, with its mature TPU pipeline, can deploy models faster, cheaper, and at scale.
And because TPUs are integrated into Google Cloud, the company can offer AI training at lower cost than AWS or Azure — a real competitive edge. “If you’re a startup building on AI,” said one venture capitalist in a CNBC interview, “you don’t just pick a cloud provider. You pick an AI stack. And Google’s stack is tighter.”
Meta’s chip team is talented. Its engineers came from Intel, AMD, and Apple. But talent doesn’t beat time. Google’s seven-year head start in custom AI silicon is now a structural advantage — one that investors can see in the numbers.
The Monetization Gap No One’s Talking About
Here’s what the earnings slides won’t tell you: Google is already charging for AI.
Businesses pay extra for AI-powered features in Workspace — smart drafting in Docs, voice summaries in Meet, automated workflows in Sheets. That’s not a future bet. That’s revenue today.
Meta? Its AI tools are mostly free to advertisers — a “value add” to keep them on the platform. No tiered pricing. No premium packages. No direct monetization.
And while Google is testing AI subscriptions for consumers via Gemini Advanced, Meta hasn’t announced any consumer AI pricing model. Its entire strategy assumes AI will make ads slightly more efficient — not open a new revenue stream.
That’s a problem when you’re asking investors to fund $35 billion in capex. Growth stocks don’t trade on efficiency. They trade on new markets, new products, new margins. Google is selling all three. Meta is selling optimization.
What This Means For You
If you’re building AI applications, this split tells you where the infrastructure momentum lies. Google’s ecosystem — from TPUs to Vertex AI to Gemini integration — is maturing fast. Its tools are stable, well-documented, and increasingly interoperable. For developers, that means less time debugging pipelines and more time shipping.
But don’t ignore Meta entirely. Its open-source AI models — Llama 3, Code Llama, smoothM4T — remain some of the best available. And if you’re working on AI for social, real-time translation, or creative tools, Meta’s research team is still pushing boundaries. Just know that its internal execution is lagging, and that could affect tooling support.
For founders, the message is clearer: investors reward AI that plugs directly into revenue. If your startup’s pitch is “AI for better engagement,” you’ll need to show a monetization path — fast. The Meta premium is gone. The Google standard is rising.
One Question the Market Hasn’t Answered
What happens if Meta’s AI finally delivers — but too late?
It’s possible. Zuckerberg has surprised before. Maybe the AI agents in WhatsApp become indispensable. Maybe generative avatars catch on. But timing is everything. By the time Meta proves its AI works, Google might already own the enterprise AI layer, the consumer AI interface, and the cloud backbone that powers both.
The race isn’t just about spending. It’s about trust, execution, and timing. And as of April 30, 2026, only one company is winning on all three.
Sources: CNBC Tech, original report
