It’s Still 1995

Author: Rex Woodbury (Digital Native) · Published: October 22, 2025 · URL: https://www.digitalnative.tech/p/its-still-1995

One-line: AI feels saturated, but on an internet timeline it is still 1995: innovation is moving fast while adoption is slow, so it is not too late, and the thing to optimise for is durable revenue rather than momentum.

Summary

Woodbury opens with Kevin Kelly’s 2014 essay arguing that “right now, today, in 2014 is the best time to start something on the internet.” Kelly’s anecdotes are about missed domains: mcdonalds.com was unclaimed as late as 1994, and ABC declined to buy abc.com. Since 2014 we have had Revolut, Ramp, Anduril, Anthropic and OpenAI. Woodbury’s claim is that the same feeling of “every venue already bulldozed” is back with agents, and just as wrong. ChatGPT is three years old, which puts us in “the first inning of a long and lucrative super-cycle. Maybe we’re top of the second.”

The case that it’s early:

• Where the companies will come from. Sequoia’s analysis says each major platform shift (cloud, mobile) produced about 20 companies doing $1B+ in revenue. Woodbury expects AI’s twenty to sit mostly at the application layer. He quotes Sequoia’s Sonya Huang and Pat Grady: competing on infrastructure or models is hopeless, but competing on apps against “corporate IT and global systems integrators… sounds pretty doable!” He calls 2025–2027 the key vintages for applied AI, the way 2009–2013 produced Uber, Lyft, Instagram, Snap, Robinhood and Coinbase.
• Where we are in the cycle. Using Carlota Perez’s ~50-year technology-revolution cycles, he places AI in the roughly decade-long Irruption Phase, even though it feels like the Frenzy Phase.
• Labour is following. A LinkedIn call for bankers and consultants wanting startup roles drew 748 responses, which he reads as a “Great Migration into tech” like dotcom, mobile and cloud.

Change takes time. Five adoption curves show how slowly even obvious shifts diffuse:

• E-commerce is only about 20% of retail, a quarter-century after Amazon’s IPO.
• Only about 60% of corporate data is in the cloud.
• About 50% of households still have cable, a decade after House of Cards.
• Apple Pay handles about 6% of eligible transactions, and mobile payments overall 15–20%.
• EVs are about 10% of new U.S. car sales.

He predicts that in 20 years “agent penetration” charts will still sit around 20%.

…but change can also be fast. He draws two consequences:

1. Beware linear anchoring. The investor who passed at $100M and balks at $500M three months later is anchoring on the past, when “if the company is growing 50% month-over-month, that $500M might look cheap pretty soon.” Amazon was up about 199x from its IPO when Kelly wrote in 2014, and it has gone up another ~15x since. OpenAI at $500B (up from $300B in March) may look cheap against its revenue forecasts.
2. Do things that don’t work today but will soon. Terrible margins may be fine in a land grab if the long-term math works.

Long-term thinking vs. short-term opportunism. He pushes back on a16z’s Bryan Kim tweet (the viral “$0 → $2M ARR in 10 days” bar) and on Hemant Taneja’s “$1M to $15M to $100M” expectation, calling the first “one of the more dangerous statements I’ve heard from a VC.” His point is that “momentum is not a moat”: Kalshi and Clay were both late bloomers, and “I’d rather a founder build a defensible product with sticky revenue and deep customer love than chase momentum out of the gate.” In his sources he credits Kyle’s essay on the same Bryan Kim statement, Momentum Is Not a Moat.

His three takeaways:

• It is still very early.
• Technological innovation outpaces technological adoption, possibly by decades.
• In a gold rush, durable moats and sticky revenue beat short-term growth.

Full text

Archived privately against link rot: ../attachments/its-still-1995/its-still-1995.md (with the 15 charts and images).

Connections

• Dr. Tokens or — How I Learned to Stop Worrying and Love the AI Bubble — the counterweight on timing: Dr. Tokens takes the bubble seriously, while Woodbury argues that adoption curves say we have barely started.
• Momentum != Moat — Kyle’s essay responding to the same Bryan Kim statement; Woodbury cites it in his sources and makes the same argument.
• Rex Woodbury — author; Digital Native and Daybreak.
• Technology Adoption Curve — the five penetration examples (e-commerce, cloud, cord-cutting, mobile payments, EVs) are the piece’s evidence that diffusion lags invention.
• Exponential Growth — the valuation-anchoring argument: linear intuitions misprice companies compounding month over month.
• Carlota Perez — the Irruption/Frenzy framing of where AI sits in the technology-revolution cycle.
• Kevin Kelly — “It’s Not Too Late” (2014) is the essay the whole piece is built on.