While the broadband business is in secular decline,
$CMCSA looks dirt cheap with a PE of 7.10, a dividend yield above 6%, and a strong balance sheet.
I don't see Comcast finding new ways to grow in the internet space. Instead, I see it playing a role in the growing demand for compute among American households.
Today, most AI workloads run in massive, centralized data centers that can be hundreds or thousands of miles away from users.
This creates latency (lag), which is the time it takes for your request to travel to the data center, get processed by an AI model, and come back.
For applications like real-time gaming, instant recommendations, or responsive AI tools, even milliseconds matter.
This issue is more important when we take into account the infrastructure needs for Physical AI (humanoid robots, robotaxis, etc.).
Comcast already has a nationwide network reaching 65 million homes and businesses, with equipment physically located close to where people live.
They have regional facilities, DOCSIS 4.0 nodes, smart amplifiers, and intelligent gateways scattered throughout neighborhoods.
With those locations, Comcast can deploy Nvidia GPUs in those facilities instead of in distant data centers, and that would boost AI responses dramatically faster.
Besides helping Comcast monetize its existing infrastructure, Comcast is already providing the American public with an alternative to large data centers.
By dedicating more compute capacity to edge sites, you'll have:
- no new mega sites
- smaller power footprint per location
- visible consumer benefit
- it supports jobs in existing communities
All the concerns over mega AI data centers go away.
This also applies to
$CHTR since they operate as a duopoly with Comcast in providing broadband to households throughout the US.
Is
$CMCSA undervalued here?
A parabolic yield surely can’t last..