AI Era: The Ultimate Deduction of the Token Supply and Demand War

Video Title: The Supply and Demand of AI Tokens | Dylan Patel Interview

Video Author: Invest Like The Best

Compiled by: Peggy, BlockBeats

Editor's Note: As AI model capabilities continue to leap forward and tools like Claude Code and Cursor are adopted on a large scale by enterprises, industry discussions are shifting from "how powerful the models are" to "how models enter production." But as AI programming, automated analysis, and data modeling become the new consensus, a more fundamental question is emerging: When execution costs are rapidly compressed, what truly becomes scarce—labor, capital, or the right to access frontier models and tokens?

Left: Host Patrick O'Shaughnessy, Right: Dylan Patel

This article is adapted from a conversation between Patrick O'Shaughnessy and Dylan Patel, founder of SemiAnalysis. Dylan, who closely tracks AI infrastructure, semiconductor supply chains, and model economics, discusses how AI is transforming enterprise organization, information services, token demand, computing supply chains, and social sentiment, starting from the explosion in his own company's Claude Code spending.

The most noteworthy aspect of this conversation isn't that a particular model has broken a benchmark again; it's that it provides a way to understand the AI economy—viewing AI as a production system that is reallocating execution capabilities, organizational efficiency, and industrial profits, rather than just a software tool upgrade.

This conversation can be broadly understood from five perspectives.

First, execution costs are being broken. In the past, ideas were not scarce; the real difficulty was turning ideas into products, systems, and deliverable services. Now, Claude Code allows non-technical people to write code, build applications, and perform data analysis. Work that once required a dedicated team for long-term maintenance is now being done by a few people with the help of models. SemiAnalysis's annualized spending on Claude Code has reached $7 million, over 25% of its payroll. This indicates that AI is no longer just a productivity tool but is becoming new production capital for enterprises.

Second, the information services industry is being rewritten first. Dylan's business essentially sells analysis, consulting, and datasets—precisely the areas most susceptible to commodification by AI. Chip reverse engineering, energy grid modeling, and macroeconomic indicator construction, which might have required a long-term team investment, can now be built into usable products by a few people in weeks. This means the pressure on information service companies isn't "whether AI will replace people," but "who can rebuild a competitor's product faster." Companies that don't adopt AI will be commodified by faster ones, while those that do must constantly raise their standards to avoid being reverse-displaced by the next wave of more efficient competitors.

Deeper still, tokens are becoming new means of production. In the past, companies bought software subscriptions, and the core question was whether the tool was good. Now, access to frontier models, rate limits, enterprise contracts, and token budgets directly determine production capacity. A stronger model doesn't necessarily mean higher costs because smarter tokens might complete higher-value tasks in fewer steps. The real competition is shifting from "who uses AI" to "who can get the strongest models and deploy the most expensive tokens on the highest-value scenarios."

This demand will continue to ripple through the entire supply chain. Soaring token usage ultimately translates into sustained pressure on GPU, CPU, memory, FPGA, PCB, copper foil, semiconductor equipment, and wafer fab capital expenditure. The "bullwhip effect" mentioned in the article is this logic: downstream, it seems like just an increase in model call demand, but upstream, it can lead to orders, capacity expansion, and price hikes amplified several times over. Profit distribution in the AI industry will therefore not stop at model companies and NVIDIA but will continue to spill over along the semiconductor and data center supply chain.

Finally, social backlash against AI may come sooner than expected. When AI truly enters workflows, public concerns about job displacement, energy consumption, data center expansion, and power concentration will rise in tandem. Dylan even predicts large-scale protests against AI within three months. For model companies, continuing to emphasize that "AI will change the world" may not alleviate anxiety; instead, it might reinforce ordinary people's sense of losing control. The AI industry next needs to prove not just its technical capabilities, but how it creates tangible, perceivable public value in the present.

Today, the core question about AI is shifting from "what models can do" to "who can get the models, how to use them, and who can capture the value they create." In this sense, the subject of this article is no longer just Claude Code, Anthropic, or any single AI company, but a structural reordering centered on productivity, capital expenditure, organizational efficiency, and social acceptance.

The following is the original content (edited for readability):

TL;DR

· The core variable for AI is shifting from "can it be done" to "is it worth doing." As execution costs plummet, the truly scarce resource becomes high-value ideas that can be amplified by models.

· Claude Code spending accounting for 25% of payroll costs is just the beginning. AI is transforming from a software tool into new production capital for enterprises.

· Competition for frontier models is no longer just about capability; it's about token access. Those who can secure the strongest models earlier and more reliably may build new competitive moats.

· The information services industry will be the first to be restructured by AI because the cost of producing data, analysis, and research is rapidly declining. Slow companies will be commodified by faster ones.

· Token demand won't slow down just because older models get cheaper. Every time a model becomes more powerful, it unlocks new high-value use cases and pushes users towards more expensive frontier models.

· The biggest change brought by AI isn't letting people work less, but letting a few people produce multiple times the output in the same time. Those who cannot create and capture value from tokens risk being locked into a "permanent underclass."

· Computing power shortages are spreading across the entire semiconductor supply chain. From GPUs, CPUs, and memory to PCBs, copper foil, and equipment manufacturers, AI demand has become a price driver for the entire industry.

· The economic value of AI is difficult to capture with traditional GDP metrics. The real question isn't just how much money model companies make, but how much "Phantom GDP" is created by the decisions, efficiency, and cascading effects generated by tokens.

Interview Transcript:

Claude Code Becomes the New Workforce

Patrick O'Shaughnessy (Host):

You once told me a fascinating story about the massive shift in your team's token usage this year. Can you tell it again? What did it teach you about what's happening in the world?

Dylan Patel (Founder, SemiAnalysis):

Last year, we thought we were heavy AI users. Everyone was using ChatGPT, everyone was using Claude, and I provided the team with whatever subscriptions they wanted. At that time, the company's spending on this was in the tens of thousands of dollars.

But this year, spending started to skyrocket. The real turning point was probably around the end of last December, with the arrival of Opus. This also includes Doug, our president, Douglas Lawler. He basically took the lead in pushing non-technical people to write code with AI. You could say he gradually brought the whole company along. Of course, the engineers were already using it, but starting in January this year, our spending clearly turned upwards, and then exploded.

We later signed an enterprise contract with Anthropic. Last time I talked to you, our annualized spending was around $5 million; now it's $7 million.

Patrick O'Shaughnessy:

And that was last week's number.

Dylan Patel:

Yes, a large part of that is just the usage volume itself. What's really interesting is that people who have never written code before are now using Claude Code, and some are spending thousands of dollars a day. But from the company's overall perspective, we're now spending $7 million a year on Claude Code, while our payroll is about $25 million. That means Claude Code spending is already over 25% of our payroll.

If this trend continues, it could even exceed 100% of payroll by the end of the year. That's a bit scary. Fortunately, I don't have to choose between "people" and "AI" right now because the company is growing fast. It's more like: I don't need to hire people as quickly, but I can spend more on AI, and it's effective, allowing the company to grow faster.

But I think other companies will eventually start facing this problem: if one person using Claude Code can do the work of 5, 10, or even 15 people, what happens next? First, layoffs might indeed be necessary; second, the use cases are currently very broad.

For example, we have a reverse engineering lab in Oregon that we've been building for a year and a half. It's full of high-end equipment, like microscopes and scanning electron microscopes. The core purpose of this lab is to reverse engineer chips, extract chip architectures, and analyze the materials used in their manufacturing. This is also one of the datasets we sell.

But analyzing this data used to be a very slow process. Now, one person on our team, spending just a few thousand dollars on Claude tokens, built an application. This app uses GPU acceleration and runs on our servers at CoreWeave. We just send it a chip image, and it automatically labels the location of each material on the image: here's copper, here's tantalum, here's germanium, here's cobalt. You can then very quickly perform finite element analysis on the entire chip stack structure, visually, with a full graphical interface and dashboard.

This person used to work at Intel, and he said that in the past, this would have been a whole team's job to build and maintain. Seeing similar things happening across the company is mind-blowing.

Another example I find particularly interesting is Malcolm. He used to be an economist at a large bank. That bank's economics department probably had 100 to 200 people. What he's building now is truly remarkable.

He connected various data sources, including FRED data, employment reports, and other datasets from different APIs. We also signed contracts with some data vendors to get API access. He pulls all this data in and runs regressions to analyze the inflationary or deflationary impact of different economic changes.

The Bureau of Labor Statistics has a whole classification of tasks, about 2,000 of them. Malcolm uses AI to assess which tasks can now be done by AI and which cannot, scoring them according to a rubric. The results show that about 3% of tasks can currently be completed by AI.

So he created an indicator to measure which things can be done by AI and what deflationary effect that will have. Output might increase, but because costs drop so dramatically, GDP could theoretically contract. He calls this "Phantom GDP."

Based on this concept, he built a whole analysis framework and a new language model benchmark containing about 2,000 evals.

Patrick O'Shaughnessy:

He did all of this alone?

Dylan Patel:

Yes, entirely alone. He told me, "Bro, this would have taken a 200-person team of economists a year to do." He's completely immersed in Claude now, saying everything has changed.

Patrick O'Shaughnessy:

As a business operator, how do you think about this? You went from almost no spending to nearly 25% of payroll, and it's still rising. At what point do you think, "Wait, should I hit the brakes? Should I control spending? Maybe we don't always need the very latest frontier model released today, like Opus 4.7, and could switch to a cheaper one?"

Dylan Patel:

Ultimately, I run an information business. We sell analysis, do consulting, and create datasets. I see no reason to believe these things won't be fully commoditized at a pretty rapid pace.

If I don't continuously improve, the first data product I sold is already facing more competition. The reason we can still sell it is that we constantly make it better and more granular. But the way we did it in 2023 isn't that different from how others are doing it now. If I don't keep raising the bar, I'll be commoditized. If I don't move fast enough, I'll lose my edge.

So the question is: yes, AI will commoditize many things, just as it's commoditizing software. But those who move fast enough, who own customer relationships, who consistently deliver excellent service and keep improving, won't shrink; they'll grow faster. The incompetent, those who do nothing, will lose.

So it's a bit of a survival problem: if I don't adopt AI, someone else will, and they will beat me.

Another simple example is the energy sector. We had a couple of energy analysts for the past year or so trying to build an energy model. It's very complex, and the energy data services market is about a $900 million market, so it's clearly a huge market I want to enter. But despite having someone on it for a year, we hadn't really broken into the energy data services business.

Then came the "Claude Code psychosis." One of our people responsible for data center energy and industrial business, Jeremy, started using Claude Code, and things changed suddenly. In three weeks, he spent a lot of money, about $6,000 a day, which is really extravagant. But he scraped every power plant in the US, every transmission line above a certain voltage level, built a map of the entire US grid from various public data sources, and integrated a lot of demand-side data.

We turned it into a dashboard where you can view and analyze power shortages and surpluses in different micro-regions of the US, along with many details. This was built in a few weeks.

Later, we showed it to some clients who already bought our data center dataset, including energy traders. They looked at it and said, "Wow, how long did this take? This is good, better than Company X." Then we looked into it and found that "Company X" had 100 people who had been working on it for ten years.

Of course, our product isn't as complete or robust as theirs yet, but in some aspects, it's already better. So I'm commoditizing these energy data service companies. But conversely, if I don't run faster, who will commoditize me?

So, from a business owner's perspective, the question isn't "am I spending a lot of money." Yes, I am. But the question is, what does this money bring me? Does it generate more revenue? If the answer is yes, then the money is worth it.

Patrick O'Shaughnessy:

Do you worry that, eventually, those who control capital, the allocators who often hire you for what you do, might say, "We have our own analysts, and they're smart too. Why don't we just do it ourselves?" If it becomes this easy, at what point does it all flow back inside investment firms? They are, after all, the ones most likely to get the biggest leverage from this data and insight.

Dylan Patel:

First, any information service business is essentially like this: the value I get from a piece of information is obviously less than the value the client gets from it.

If I sell you information for $1, you're willing to pay that $1 because you know it helps you make a decision that earns you more than $1. That is, you get an arbitrage opportunity. You make more money from me than I make selling the information.

Investment funds themselves certainly have their own information service capabilities. Especially firms like Jane Street, Citadel—they are very granular and deep on data. Yet these institutions still buy our data, continue to buy it, and our collaboration is growing.

I think there's a certain "it factor." We move faster, are more agile, have a smaller team, and focus on a very specific niche: AI infrastructure, and the massive changes it triggers, including AI, the token economy, and the whole ecosystem. We can see the direction earlier and build things faster.

So, investment professionals will certainly try to do some of what we do themselves. But more often, they'll just buy our data and build on top of it. For them, buying our data and building up from it is usually cheaper than starting from scratch. Of course, eventually, some will try to do it themselves.

Tokens Become New Means of Production

Patrick O'Shaughnessy:

I feel like every time I talk to you, I end up back at the same question: token supply and demand. It's the most interesting thing in the world to me right now. Your own experiences—what have they taught you about the demand side? Has your judgment on token demand changed after feeling this so personally?

Dylan Patel:

If we step back and look at the macro picture, Anthropic's ARR might have grown from $9 billion to around $350 billion, $400 billion. By the time this episode airs, maybe it's $400 to $450 billion.

But their compute hasn't grown at the same rate. If you do the math and assume they haven't reduced compute for R&D—which they obviously haven't, as they keep releasing new models like Metis, Opus 4, Opus 4.7—it means one thing: even if all the new compute went to inference, their gross margin floor would be around 72%.

In reality, some of the new compute likely went to R&D, so their actual gross margin might be even higher than 72%. Remember, earlier this year some