Big Tech doubles down on AI infrastructure while markets debate the “AI bubble”

Here we go again.

Meta just signed a multibillion-dollar agreement with AMD to secure custom AI chips for future infrastructure expansion, while Microsoft continues scaling its global internet ambitions through its partnership with SpaceX’s Starlink network.

Taken together, these moves suggest something important:

Major tech companies are no longer experimenting with AI, they’re committing capital at infrastructure scale.

Meta’s deal reportedly targets up to 6GW of custom GPU capacity starting in 2026, signaling long-term demand for compute power beyond current AI hype cycles.

At the same time, Microsoft’s Starlink collaboration expands connectivity to underserved regions, potentially onboarding hundreds of millions of new users into cloud and AI ecosystems.

From a market perspective, infrastructure investments like these often precede revenue expansion by several years.

Which raises an interesting question for investors:

If AI spending continues accelerating despite recession fears and valuation concerns, are we actually seeing the early phase of a productivity cycle rather than a speculative bubble?

Personally, I’ve noticed that markets tend to react extremely fast to AI-related announcements. Some recent equity moves happened before most retail investors even understood the catalyst, which shows how positioning around macro narratives increasingly matters. I even captured a rapid gain because I was already positioned long on Bitget Stock Futures when Jack Dorsey announced Block’s 40% workforce reduction tied to its shift toward an AI-native operating model, which triggered an immediate 20–25% after-hours surge before most market participants fully priced in the news.

Curious how others see this:
Are AI infrastructure investments still underpriced, or are markets already pricing peak optimism?