AI has forced me to rethink not only how I read companies, but also how I read markets.
What began as industry analysis gradually led me back to skills I once used as a financial analyst.
This essay is a reflection on why that happened.

I never expected that one day I would write an essay like this.

For me, industry analysis has always had a certain kind of clarity. You can set stock prices aside for a while and focus on how technology evolves, how supply chains shift, how companies compete, how product logic changes, and which signals truly matter even if the market does not notice them right away. In many cases, even when stock prices swing sharply, companies still have enough cash flow, a strong enough business base, and enough investment capacity to keep moving forward with what they want to do. Markets matter, of course, but they do not necessarily reshape the direction of an industry every single day.

So when I returned to industry analysis in recent years, I was no longer watching stocks every day the way I did earlier in my career as a buy-side analyst. More than short term market moves, I wanted to understand what was really happening in the industry and what kind of strategic logic was sitting behind corporate decisions.

But once my work led me to look seriously at AI, I began to realize that this industry was different. Over time, I came to see that AI was forcing me to recover some of the skills I had once relied on as a financial analyst. Not because I suddenly wanted to switch back to making investment calls, and not because I had stopped believing in the value of industry analysis itself, but because the way the AI industry operates has become increasingly difficult to understand without bringing capital markets back into the picture.

This is not just a race in technology. It is not just a race in products. It is also a race for expansion that requires the continuing consent of capital markets.

In the past, industry analysis did not require this kind of market lens

When I worked on many technology sectors in the past, I did not need to tie stock prices, market sentiment, and corporate actions together so closely. The reason was simple. Capital markets were always there, of course, but in many mature industries, or for many established companies, investment remained broadly within a range that firms could absorb, plan for, and support through their existing cash flow and balance sheets.

In other words, even when markets turned volatile, companies still retained a meaningful degree of autonomy. They could move early because they believed in a direction. They could make long term investments before the market fully understood them. The market might question those decisions. It might even push back. But those companies were not necessarily forced to stop right away. In some cases, they could move first and explain later.

In that kind of industry, the stock market and the industry itself were never completely separate, but there was still some distance between them. The market could undervalue you, and it could overvalue you, but the basic logic driving the industry forward did not necessarily change at once because of that.

That is why, in the past, I could stay relatively focused on the industry itself. I wanted to understand technological progress, shifts in the supply chain, customer adoption, product innovation, and the real sources of competitive advantage between companies. As for market fluctuations, I could often place them one layer lower in my analysis.

AI made me realize that this was no longer enough.

The scale of AI investment has changed the method of analysis

What truly changed the way I observe this sector was not any single company, and not any one earnings report. It was the scale of AI investment itself.

When corporate spending on AI was still limited to new products, new models, or narrower research efforts, it could still be understood as a normal extension of technology investment. But once it began to take the form of massive data center construction, long term compute commitments, large scale purchases of GPUs and networking equipment, early moves to secure power and land, multi year infrastructure planning, and a nearly nonstop chase from one model generation to the next, it stopped looking like research spending in the ordinary sense.

It became a form of expansion defined by extreme capital intensity, very large upfront commitments, and a payback path that is still far from clear.

At that point, it became difficult for me to look only at what a company wanted to do without also asking whether the market was willing to let it continue. Investment at this scale cannot remain untouched by capital markets.

A company may have vision, strong technology, capable management, and a clear long term objective. But if the market starts questioning the return on those investments, the size of capital spending, or whether commercialization can keep pace with the scale of investment, those doubts do not remain confined to stock charts. Over time, they begin to shape the company’s room to act.

Will management still feel comfortable expanding aggressively. Will investors remain willing to tolerate pressure on near term profits. Will debt costs and capital flexibility begin to shift. Will the market still grant these companies enough time for AI returns to mature.

In many industries, these were not questions that had to sit at the center of analysis every day. In AI, they have become core conditions that are too important to ignore.

I gradually realized that the market is not just an observer

It was also through this process that I slowly came to see that, in AI, the market is no longer just standing on the sidelines evaluating companies. It is increasingly becoming one of the external forces that helps govern whether this industry can keep expanding.

That may sound a little strong, but it comes very close to how I have come to see things. In the past, the market mostly watched what companies were doing. Now it also shapes how much they feel able to do. The market used to act more like a scorekeeper. Now it is starting to look more like a setter of conditions.

If a company’s AI story is still accepted by the market, it has more room to expand data centers, move early on the next platform cycle, absorb pressure on short term profits, and keep investing in directions that have not yet been fully proven. But if the market begins to lose patience with that story, then even if the company still believes the direction is right, it may no longer be able to keep pushing forward without meaningful cost.

In other words, stock prices are no longer just an outcome. Valuation, financing capacity, market patience, and the cost of capital once sat more clearly on the financial side of analysis. Now, to some extent, they have become part of the operating logic of the AI industry itself.

That is why I eventually felt that I could not avoid looking at stocks again. Not because I wanted to turn industry analysis into investment advice, and not because I believe the market is always right, but because if I do not understand how the market prices AI, how much patience it has, and how its expectations are changing, I will likely miss one of the most important parts of the industry’s reality.

Why I began to pay closer attention to tech narratives

I became increasingly interested in tech narratives not because I had suddenly become interested in stories for their own sake, but because I gradually realized that the market rarely understands an AI company simply by looking at what it has actually done. More often, it relies on a narrative framework that translates technology, investment, and future returns into a story the market can accept.

When I used to do industry analysis, I cared more about what a company was actually doing. Was the technology mature. Did the product truly work. Was the supply chain keeping up. Was the strategy internally consistent. Those questions still matter today, and I still believe they are among the most important ones.

But at this stage of AI, I have gradually come to see that, important as they are, they are no longer enough to explain how far a company can go.

The market does not simply look at what a company has done and arrive naturally at a matching conclusion. It uses its own language, its own expectations, and its own narrative frames to interpret the same company. It decides whether that company looks more like an infrastructure builder, a platform integrator, a future profit engine, or a risky bet with oversized investment and unclear payback.

And companies are not just passive recipients of those interpretations. They actively shape stories of their own, trying to persuade the market that today’s spending is reasonable and that future returns are worth waiting for. They reframe their positioning, adjust their language, amplify certain narratives, soften certain risks, and try to translate their investment logic into terms the market can accept.

I came to realize that the gap between these two sides is itself part of AI industry analysis. Of course, this kind of gap is not unique to AI. Many technology sectors have always contained a difference between how the market interprets a company and how the company wants to be understood. But in the AI era, that gap has become much more important. When investment becomes enormous and the payback period remains uncertain, narrative stops being just a companion to valuation. It becomes more like a tool for securing time, capital, and market patience.

From that perspective, the reason I chose to write about AI tech narratives through an independent research project was not that I wanted to move away from industry analysis. It was that I came to see that, in the AI era, narrative itself has become part of the conditions under which the industry operates.

If we do not understand how the market interprets a company, and how the company actively designs its own story, it becomes difficult to fully understand why that company can expand, why it comes under doubt, or why at a certain moment the market suddenly runs out of patience.

AI has brought industry analysis and financial analysis back together

For me, this has been a very interesting shift. I began my career as an industry analyst, later moved into buy-side analysis, and spent a period working much closer to the market. When I eventually returned to industry research, I had a very natural instinct to screen out market noise and go back to the industry itself. There was nothing wrong with that, and in many cases it was still necessary. If a researcher lets the market pull too strongly on their thinking too early, it becomes easier to lose sight of the signals that matter over the long run.

But AI seems to have brought these two skill sets back into contact. It has made me realize that once an industry reaches a certain stage, industry analysis and financial analysis can no longer be kept so far apart. Capital markets do not just reflect a narrative. They also affect whether that narrative can continue to hold. Companies are not moving forward on technology and products alone. As they expand, they also need continued market tolerance, support, and time.

That feels very different from the rhythm of industry analysis I was used to. In the past, I could focus on whether a company’s technological direction was right, whether product execution was on track, and whether its position in the supply chain was secure. Now I also have to think one layer further and ask whether the market is willing to support that company all the way through. In other words, I used to treat capital markets as an important but external variable. Now that has become harder to do, because markets increasingly look like part of the industry system itself.

In a way, AI has pulled me back toward a place I thought I had already moved beyond. It has reminded me that understanding a company and understanding the market are not always two opposing skill sets. Sometimes they are simply two ways of reading the same question at different levels.

This also changed the way I look at companies

This shift in method also changed what I pay attention to when I look at companies. I now care more than I used to about questions that would not always have been at the front of my mind. Which AI strategies depend more heavily on sustained, large scale capital spending. Which narratives are more vulnerable to changes in market expectations. Which companies may have strong technology but still need more time to turn that strength into stable commercialization. Which ones have a more complete platform and product structure, making it easier to turn AI from an expensive capability into a service that can actually be delivered, managed, and adopted by enterprises.

I also began to care more about the relationship between a company and the market. Not just what the company itself is doing, but how the market understands those actions, and whether that understanding may in turn shape the company’s future capital allocation and ability to expand.

That is why it has become harder for me to treat stock prices as mere emotional noise. Of course, markets can overheat, and they can also become overly pessimistic. Short term distortions will always exist. But at this stage of AI, the market is not always just a source of misunderstanding. At times, it can also act as an accelerator or a constraint.

That means researchers looking at AI can no longer ask only whether the technology is right. They also have to ask whether the capital support behind that technology can be sustained. They cannot ask only whether a product can work. They also have to ask whether the market is willing to give it enough time and resources to become real. They cannot ask only what a company wants to do. They also have to ask whether the company has the capacity to bear the cost of that direction.

AI forced me to reconnect these two skill sets

If I had to name the biggest change in me during this period, it would be this. I finally had to admit that if I want to understand AI in a complete way today, I cannot rely on just one kind of training. That does not mean I have gone back to being a financial analyst. At least, I am not looking at things from that identity.

But AI has undeniably forced me to recover part of an older skill set. Not so I can make investment judgments more quickly, but so I can understand a new reality. Once technological development enters a phase of extreme capital intensity, the market no longer serves only as a mirror reflecting the industry. It starts to become part of the industry itself.

That realization has felt both fresh and a little surprising to me. I had assumed that once I returned to industry research, I would gradually move farther away from the market. The opposite turned out to be true. It was precisely because I wanted to understand AI more seriously that I realized I could not avoid understanding the market again.

Seen from that angle, this feels less like a change in identity and more like a methodological correction I could no longer avoid. AI has not made industry analysis less valuable. It has simply made me realize that if industry analysis still hopes to remain complete today, it has to bring capital markets back into the frame. And the market matters not only because it reflects price, but because through narrative, expectation, and tolerance, it also helps determine which companies are given the chance to carry their story forward.

Conclusion

Through this process, I slowly came to understand why analyzing AI brought back some of the habits and instincts I once had as a financial analyst.

It was not because I had changed my belief in research, and not because I wanted to turn every industry question into a stock price question. It was because the AI industry itself has become very difficult to understand in a complete way without understanding capital markets.

It requires more capital, moves at a faster pace, faces denser competition, and offers a payback timeline that is not always clear. That has made the market more than a judge scoring things after the fact. To some extent, it now participates in the game itself by setting conditions and allocating time. Perhaps that is one of the new realities of the AI era.

If industry analysis in the past could assume that the most important question was what a company was actually doing, then AI analysis today may need to add one more line. What matters is not only what a company wants to do, but also how the market understands that ambition, and whether the market is willing to let the company keep going.

I think this may be one of the new questions that the AI era has left for industry researchers.