And why being right about the technology still doesn’t mean being right about the price
- I’ve seen this before: Cisco was right about the internet and still took 25 years for its stock to recover
- Seven technologies — LLMs, agentic AI, compute, robotics, blockchain, energy, quantum — are advancing simultaneously and accelerating one another
- Unlike past revolutions, there’s no rail to lay or fiber to bury: distribution to five billion people exists
I started my investing career in June of 1999. That gave me nine months of watching technology stocks go nearly straight up before the bubble burst — and it’s the kind of windmill you only fight once, but never stop seeing.
I’m seeing it again.
They say the windmills you fight hardest are the ones you faced early in your career, and the dot-com bubble is mine. Consider Cisco Systems. In March 2000, Cisco became the most valuable company on Earth — $80 per share, $500 billion market cap.[1] The thesis was right: the internet did change everything, and Cisco was at the center of it. Revenue grew from $19 billion to nearly $60 billion over the next 20 years.[2] The stock fell more than 80% when the bubble burst, and it took until December 2025 — a quarter of a century — for the share price to recover.[1] The lesson was never that the technology was wrong. It was that valuation matters, even when you are right about the future.
The parallel is worth considering. This time, the technology case may be the strongest one I’ve seen in more than twenty-five years of doing this.
When ChatGPT launched in late 2022, it is estimated that it reached 100 million users in just 60 days — the fastest adoption of any consumer application in history.[3] Statista estimates that it took the telephone 50 years and the mobile phone 12 years to reach half that number.[4]
That moment, as remarkable as it was, is likely being surpassed by what is happening right now. Marc Andreessen, co-founder of Andreessen Horowitz and Netscape, calls it “the biggest technological revolution of my life … clearly bigger than the internet.”[5] Nicolai Tangen, who runs Norway’s $2 trillion sovereign wealth fund, described the advancement curve of today’s technology as “nearly vertical.”[6]
Three things, in our view, set this moment apart.
- First, the sheer breadth. Each of these would be a major story on its own: Large language models. Agentic artificial intelligence. A generational leap in compute power. Blockchain and the tokenization of real-world assets. Humanoid robotics. A massive expansion in energy infrastructure. Quantum computing. That all seven are advancing simultaneously — it is hard to find a historical parallel.Of these, two strike me as having the most immediate portfolio relevance. The first is agentic AI — the shift Jensen Huang described as moving from information to real work. Where most people today still use AI as a glorified search engine, agentic systems can be handed a complex project, split it across multiple AI workers running simultaneously, and hand you something you can use. This is already happening in software development, legal research, and financial analysis, and the companies building the infrastructure for it are worth watching.The second is humanoid robotics. The feedback loop driving self-driving vehicles — where AI observes the real world, uploads to a central model, and pushes improvements back to every unit on the road — is now being applied to physical robots. The technology is not quite there yet on fine motor tasks and speed, but the rate of improvement is what matters: every hour these systems operate makes them better, and that compounding tends to surprise people.We’re wired to think linearly, which is why a force that compounds — growing by percentage rather than by a fixed amount each period — is so easy to underestimate until you see the changes up close, in the people living through it.A friend is a radiologist in his 70s. He expected to be retired by now, but instead, he is busier than ever — working from home, aided by AI, and still in demand because someone credible needs to sign off on what the model produces. He told me that the AI is finding things he would have missed, and he is catching things the AI gets wrong. The combination is better than either one alone.
That is not the story most people are telling about AI and jobs, but I think it is closer to what is happening, at least for now. In fields like healthcare, financial services, law, and engineering, the clearest effect of AI is not that it replaces professionals. It is that it helps them do more and often better.
Some roles may shrink, and some tasks may disappear. But the more immediate reality is that AI is amplifying the capabilities of good professionals rather than eliminating the need for them. That shift alone is transformative, and the companies enabling it are worth paying attention to.
- Second, the convergence. These technologies are not advancing in isolation. They are building on top of one another, accelerating each other. More powerful chips make AI models smarter; smarter models design better chips and write better code; better code accelerates everything else. Jensen Huang, CEO of NVIDIA, recently said, “computation has increased 10,000x in just two years”.[7] Andreessen contends the cost of intelligence is falling faster than Moore’s Law, even as capability rises.[5] To put that in concrete terms: NVIDIA’s Hopper chip was the hardware that powered the original ChatGPT moment. Its successor, Blackwell, is already delivering dramatically higher performance, while NVIDIA’s next-generation Rubin architecture, expected in the coming years, promises another leap forward.
- Third, the lack of constraint. Prior technological revolutions were throttled by physical infrastructure. You had to lay rail for the railroads, string wire for the telephone, and bury fiber for the internet. This revolution rides on an already installed internet that reaches an estimated five billion people.[8] The moment a new model is trained, it is available to anyone with a smartphone. The only meaningful constraint is energy, a solvable engineering problem that many companies are focused on.
A word on where the models stand, because I think most people underestimate them. The common critique is that AI makes mistakes, and it does. But humans make mistakes too, and we rarely apply the same base rate scrutiny to human judgment that we apply to machines. Michael Mauboussin has written about this tendency — in Think Twice, he argues that we consistently judge performance against an imagined ideal rather than against the realistic alternative — and AI is a case in point. The question is not whether AI is perfect, but whether it outperforms the human process it is replacing, and the evidence is mounting.
It is hard to imagine a more exciting time to be alive from a technological perspective.
Why valuation still matters
That is the hardest part of this. Being right about the technology does not mean you are right about the price — and it is easy to confuse the two.
The trouble with windmills is that you only know after the fact whether they were imaginary or real.
At the moment, our concern is not that long-term investors pay too much attention to this revolution. It is that they pay too little. But every investor should be wary. While we are very excited about the impact of the exponential convergence of these technologies, it is hard to ignore the similarities to the technology bubble of the early 2000s. For clients wondering what this means for their own portfolios — whether they are over- or underexposed, and how to think about valuation discipline in a rapidly changing landscape — that is exactly the kind of conversation we are built for. We’d welcome it.
Sources
[1] CNBC — cnbc.com (Cisco $500B market cap, $80 share price, March 2000; 88% decline; 25-year recovery, December 2025)
[2] Companies Market Cap, Cisco Revenue History — companiesmarketcap.com
[3] UBS via Reuters — reuters.com (ChatGPT reached 100 million users in two months, February 2023)
[4] Statista / Visual Capitalist — statista.com (Adoption timelines: telephone 50 yrs, mobile 12 yrs)
[5] Marc Andreessen, a16z Podcast — “Marc Andreessen’s 2026 Outlook: AI Timelines, US vs. China, and The Price of AI”
[6] Nicolai Tangen, NBIM — youtube.com (“Nearly vertical” AI growth curve, March 2026)
[7] Jensen Huang, All-In Podcast — youtube.com (10,000x computation increase in two years, March 2026)
[8] ITU / World Bank — itu.int (Approximately 5 billion internet users worldwide, 2024)
Disclosures
Gryphon Wealth is a fee-only fiduciary registered investment adviser regulated by the SEC. Registration does not imply any level of skill or training. This article is for informational purposes only and does not constitute investment, tax, or legal advice. Past performance does not guarantee future results.
This article was created with the assistance of artificial intelligence as part of our research and drafting process. Every piece of content is reviewed, edited, and approved by the Gryphon Wealth team before publication to ensure accuracy, clarity, and alignment with our values. The final perspective shared here is our own.