One of the hardest things about investing in artificial intelligence is that nobody truly knows the future.
That may sound obvious, but it is worth emphasizing as we close the first half of 2026, a period in which AI has dominated market alpha. At the same time, the AI cycle has created an unusual level of conviction on both sides. Bulls are convinced AI will change everything. Bears are convinced the market has already gone too far. Both sides can build persuasive arguments. Both sides can point to history. Both sides can sound intelligent.
But the defining feature of this cycle is that the future is now moving faster than the human brain’s ability to model it. Artificial intelligence is advancing at a pace where human intelligence is struggling to keep up.
AI is not a normal technology cycle. The products are improving quickly. The usage curves are changing quickly. The infrastructure requirements are changing quickly. The business models are still forming. The winners and losers are still being discovered in real time. Even the companies building the technology are still discovering how large the opportunity may become and how quickly it may arrive.
That is why humility matters so much.
Stanley Druckenmiller once said, “I’ve been humbled many times in my career, and I’m sure I’ll be humbled many times in the future.”
That is the right mindset for this AI cycle. If one of the greatest investors of all time can openly admit that markets have humbled him repeatedly, then the rest of us should be careful about pretending we can forecast the future with certainty. The goal is not to prove we know how the AI cycle ends. The goal is to stay humble enough to let earnings, price, and the facts on the ground tell us when the evidence has changed.
Druckenmiller has also said, “The best economist I know is the inside of the stock market.” That line is especially relevant today. Market internals often begin to warn before the official data or consensus narrative catches up. Until they do, using a narrative to call the top in a transformational technology cycle is not humility. It is prediction disguised as discipline.
When the future is uncertain, investors often fall back on stories. They decide something is a bubble because it looks like prior bubbles. Or they decide something is unstoppable because the technology feels transformational. But in a rapidly changing environment, narratives are not enough. The market needs evidence.
That evidence comes from two places: earnings and price.
Earnings tell us whether the story is becoming reality. Price tells us whether the market is confirming or rejecting that reality. When both are moving in the same direction, fighting them becomes an uphill battle.
This is especially true in AI because earnings are not moving slowly. Revenue, margins, backlog, capex, token usage, data center demand, power constraints, networking needs, memory demand, and software adoption are all changing at speeds that make traditional forecasting difficult. In a normal cycle, investors may have time to debate whether expectations are too high. In this cycle, the numbers themselves are being revised before the debate is finished.
Valuation still matters. Stocks can still go down. Not every AI company deserves the benefit of the doubt. But if earnings are accelerating and price momentum remains strong, investors should be careful about fighting that combination simply because a stock feels expensive or a narrative feels crowded.
There is a difference between being early and being right.
A stock can be overvalued for a long time if earnings continue to surprise to the upside. A theme can look crowded for months or years if the fundamentals keep improving. A market can appear obvious in hindsight only after the price trend finally breaks. This is why calling the top in a powerful earnings cycle is so difficult. You are not just betting against valuation. You are betting against revisions, liquidity, positioning, management commentary, and price confirmation all at once.
That is a hard battle to win.
This is also why I have tried to organize my work around Signal, Alpha, and Agency. The goal is to help people navigate this fast-moving roller coaster of an AI cycle without getting thrown around by every headline, narrative, or market swing.
Signal is the first job. In a world filled with noise, narratives, and biases, investors need a way to separate what is happening from what people are saying is happening. The AI cycle produces more noise than almost any theme I have ever seen. Every day there is a new claim that the bubble is over, that China has changed the economics, that capex will collapse, that the hyperscalers are spending recklessly, or that the entire trade is about to unwind.
Some of those concerns may eventually prove right. Many will not. Signal exists to filter, not to cheerlead. It separates the bear case from the facts, the bull case from the evidence, and the narrative from the data. What are the facts? What is the narrative? What is changing in the actual data? What is only changing in sentiment? What is an earnings fact, and what is just a market bias dressed up as analysis?
That distinction matters because narratives are cheap. Earnings are expensive. A narrative can be created in a day and go viral in today’s world of real-time information access. Earnings take customers, products, infrastructure, capital, supply chains, and demand. When the narrative says one thing but earnings and price say another, I want to respect the evidence first.
Alpha is the second job. Once the signal is clear, the next question is what stage of the AI cycle are we in, who should benefit and where are the earnings and price action are actually working. It is not enough to say “AI” as one broad theme. The market is constantly separating the companies benefiting from the cycle from the companies merely associated with it.
This is where earnings and charts matter so much. The companies that deserve attention are the ones where the numbers are moving and the price is confirming. That does not make them risk-free. It means they have earned the right to stay on the field. In a market moving this quickly, there is no need to fight companies where earnings revisions are positive, management commentary is strong, demand is visible, and the chart remains in an uptrend.
To close out the first half, out of the 100 names in my thematic portfolio, 83% are above their 200-day moving average, 84% have their 50-day moving average rising, and 86% have their 200-day moving average rising. This is an uptrend in the agentic AI buildout.
If that changes, we should change. But until then, the burden of proof is on the argument that says the market is wrong.
This is one of the main reasons I want Alpha to focus on the companies whose earnings and price are working. In an uncertain world, the market gives us clues. A company with accelerating earnings and strong price momentum is telling us something different from a company with a good story but deteriorating numbers. One is evidence. The other is hope.
Agency is the third job, and it may be the most important over time. AI is both an investment theme and a tool people need to use. This is where the month of June was a very satisfying month for me. The number of subscribers who went through the fight to build the Jensen Huang Knowledge Brain and then sent me emails was awesome. Hearing how that spread to their children made it even more satisfying. Agency gives people both an empowered feeling and the ability to look through the noise from experience.
One of the biggest risks in this cycle is that people form opinions about AI without personally experiencing how powerful it is becoming. It is easy to see by just going back in history to see if their views have not changed since the release of ChatGPT. If someone only reads headlines without checking the bias, they are more vulnerable to narratives. If they use the tools, build workflows, test agents, compare models, and see what AI can do in their own life and business, they develop a better understanding of the technology’s practical power.
That does not make them immune from mistakes. But it makes them harder to fool.
Agency is about helping people move from passive observer to active participant. Along the AI progress the last year, I have supplied, how-to videos, prompts, examples of things to build like the Knowledge Brain and weekly AI built spreadsheets to help with the price and earnings trends. The more someone uses these tools and ultimately AI, the more they understand why demand is rising, why token usage matters, why infrastructure is constrained, why productivity gains may be real, and why the market is struggling to price the opportunity. They also become better equipped to challenge hype because they can distinguish between what the technology can actually do and what is still marketing.
That is the combination I wanted to bring to people this year.
Signal helps people filter facts from noise.
Alpha helps people identify where the earnings and price action are confirming the opportunity.
Agency helps people use AI personally, so they understand the technology from the inside rather than reacting emotionally to every headline.
Together, those three layers create a process for navigating a future nobody can predict.
The better approach is evidence over prediction. If the earnings begin to weaken, that matters. If the price action begins to break, that matters. If both weaken together, that matters even more. But until one or both of those signals change, there is no need to force a bearish conclusion simply because the move has been large.
In fast-moving technology cycles, the market often looks expensive before the full earnings power is visible. That is the tension investors face today. If AI is just another capex bubble, then today’s valuations may prove excessive. But if AI is creating a new productivity layer, a new compute economy, and a new demand curve for infrastructure, then the earnings base itself may be far larger than traditional models assume.
The problem is that we cannot know that with certainty in advance.
That is why price and earnings become so important. They are not perfect signals, but they are the best real-time feedback mechanisms we have. Earnings show whether customers are actually spending. Price shows whether investors are continuing to reward that spending. When both are positive, the market is telling you that the burden of proof remains on the bear case.
This is where many investors get trapped. They begin with a reasonable concern: margins may peak, capex may overshoot, competition may increase, returns on investment may disappoint, or the stocks may already discount too much good news. Those are valid concerns. But then they turn those concerns into a conclusion before the evidence confirms it.
That is dangerous.
In a cycle this powerful, the question should not be, “Can I imagine a reason this ends badly?” Of course you can. Every cycle ends eventually. The better question is, “Has the evidence started to confirm that the cycle is ending?”
If earnings are still rising, revisions are still positive, and price trends are still intact, the answer is probably no.
This is not blind bullishness. It is respect for the trend until the trend changes. Markets are discounting mechanisms, but they are also learning mechanisms. In AI, the market is learning alongside everyone else. Every earnings report, every product release, every capex update, every token usage figure, every data center announcement, every supply constraint, and every management comment adds new information.
When the information changes, the view should change. But the view should not change simply because the move feels uncomfortable.
The history of markets is filled with examples of investors who were intellectually correct but financially early. They identified excess, but the stocks kept rising. They saw valuation risk, but earnings kept compounding. They understood the eventual problem, but they underestimated the duration of the cycle. In practice, timing matters. Being right too early can look exactly like being wrong.
That is why waiting for confirmation is not weakness. It is discipline.
For AI investors, the key is to separate prediction from process. Prediction says, “I know how this ends.” Process says, “I will follow the evidence as it evolves.” Prediction creates ego. Process creates adaptability. In a world where the technology is changing rapidly and nobody knows the future, adaptability is far more valuable than certainty.
That is the purpose of Signal, Alpha, and Agency.
Signal keeps us grounded in facts.
Alpha keeps us focused on where the market is rewarding actual earnings power.
Agency keeps us personally connected to the technology so we are not dependent on headlines, narratives, or the biases of others.
The market does not require us to know the final destination of AI today. It requires us to observe whether the path is still improving or starting to deteriorate.
Right now, the most important question is whether earnings and price are already signaling disappointment. If they are not, fighting both is unnecessary.
There will be a time to become more cautious. There always is. But that time should be driven by evidence: slowing earnings, weaker revisions, failed breakouts, broken leadership, deteriorating breadth, or management teams beginning to guide down expectations. Until then, the discipline is to let the market and the fundamentals speak first.
In a rapidly changing AI world, nobody gets paid for pretending to know the future.
We get paid for listening when the evidence changes. We will listen together.
