History shows that major technological shifts like the internet and AI require a fundamental re-architecting of everything from silicon and networking up to software. The industry repeatedly forgets this lesson, mistakenly declaring parts of the stack, like hardware, as commoditized right before the next wave hits.

The proliferation of sensors, especially cameras, will generate massive amounts of video data. This data must be uploaded to cloud AI models for processing, making robust upstream bandwidth—not just downstream—the critical new infrastructure bottleneck and a significant opportunity for telecom companies.

Starlink's satellite beams are too broad to effectively serve dense cities. Its business model is complementary to ground-based cellular, focusing on rural and underserved areas where building fiber or cell towers is economically inefficient.

Unlike 4G/5G revolutions driven by consumer video, 6G will be defined by its utility for enterprise AI applications. Key advancements will be in managing network performance, reducing latency, and adding security layers crucial for business, rather than just increasing consumer bandwidth.

When a new technology stack like AI emerges, the infrastructure layer (chips, networking) inflects first and has the most identifiable winners. Sacerdote argues the application and model layers are riskier and less predictable, similar to the early, chaotic days of internet search engines before Google's dominance.

Unlike the dot-com bubble's finite need for fiber optic cables, the demand for AI is infinite because it's about solving an endless stream of problems. This suggests the current infrastructure spending cycle is fundamentally different and more sustainable than previous tech booms.

The massive capital expenditure in AI infrastructure is analogous to the fiber optic cable buildout during the dot-com bubble. While eventually beneficial to the economy, it may create about a decade of excess, dormant infrastructure before traffic and use cases catch up, posing a risk to equity valuations.

The intense demand for throughput and low latency from AI workloads is forcing a rapid migration to higher speeds (from 100G to over 1.6T). This has drastically compressed the typical five-year hardware refresh cycle down to just 12-18 months, a pace previously unheard of in networking.

The next wave of data growth will be driven by countless sensors (like cameras) sending video upstream for AI processing. This requires a fundamental shift to symmetrical networks, like fiber, that have robust upstream capacity.