SemiAnalysis Analyst Breaks Down the Current Pullback: Semiconductors Are Paying Their Dues, But the Cycle Isn't Over Yet

Compiled & Translated by: Deep Tide TechFlow

Guest: Doug O'Loughlin, SemiAnalysis Analyst (Former Founder of Fabricated Knowledge)

Host: Dylan Patel, Founder of SemiAnalysis

Podcast Source: SemiAnalysis Weekly

Original Title: Ep. 022 - Market Drawdown, Historic Bubbles, Funding The Buildout, AI Politics (Doug is Back)

Air Date: July 29, 2026

Conflict of Interest Statement: Doug O'Loughlin and Dylan Patel are employees of SemiAnalysis, a paid research firm for the semiconductor industry whose business model depends on industry health. The following content faithfully presents the original dialogue and does not constitute investment advice.

Key Takeaways

Doug O'Loughlin makes a long-awaited return to SemiAnalysis Weekly, coinciding with a sharp pullback in the semiconductor sector following its "best first half ever." South Korea's KOSPI has fallen 40%, retail investors using 2x leverage have been wiped out, and SK Hynix missed expectations due to slower price increases from shifting towards more LTAs. Doug compares the current situation to the Taiwan bubble of the 1980s, noting that the behavioral patterns of bubbles are highly similar, although fundamentals remain healthy.

The two engage in a heated debate about "just how big is AI demand?" Dylan draws on SemiAnalysis's own experience: since deploying coding agents, the company's AI spending has increased 100x, users expanded from 9 to 90 people, and per-user usage grew 10x. Doug doesn't deny strong demand but raises a core concern: while the supply side can be calculated, the demand side is a "trillion-dollar question" with no answer. More critically, scaling laws require doubling chips, but physical and institutional bottlenecks like electricians, capital, and permits cannot be doubled synchronously. Hyperscalers have already issued $450B in debt this year, sourced from pensions and annuities, yet the pension pool itself is shrinking.

Key Insights Summary

On Market Drawdown

"Through the end of Q2, it was the best performance in semiconductor history. Then we started paying the piper. The faster things rise, the harder gravity pulls them down."

"Koreans have a 20-year track record: they always buy at the top. Banks in 2007, SaaS in 2021, this time they yolo'd themselves."

"KOSPI drops 40%, people with 2x leverage get wiped out. Then it's a self-fulfilling spiral: everyone watches their accounts shrink, decides to sell, which exacerbates the decline."

On the Memory Cycle

"SK Hynix shifted towards more LTAs, price increases slowed from 3x to 30-50%. The financial world is completely broken, only looking at rate of change. As soon as the second derivative comes down, they think the cycle is over."

"The semiconductor playbook is always the same: during shortages, everyone double orders. Factories see demand and ramp up capacity crazily. Then demand sneezes, supply is still ramping, utilization drops from 100% to 50%, and they have to slash prices."

On AI Demand

"The demand curve is the trillion-dollar question. The supply curve is relatively understandable, but whether demand is 10x or 100x, nobody knows."

"SemiAnalysis itself is a case study: after deploying coding agents, 9 technical users became 90 company-wide users, and each person's token usage went up 10x. The company's AI spending increased 100x."

On Supply Chain Bottlenecks

"The US is short 100,000 electricians. A mid-level electrician makes $250k a year, those willing to work overtime can get $400k-$500k. Someone is using a Cessna to fly electricians to remote job sites."

"Hyperscalers have issued $450 billion in debt this year, second only to the US government and China's borrowing. This money comes from pensions and annuities, but the pension pool isn't doubling."

"TSMC directly and indirectly accounts for 20% of Taiwan's GDP. To double again, Taiwan would need to have more babies just to get enough workers."

On AI Politics

"AI is less unpopular than ice cream, less unpopular than politicians. This isn't priced in. In the midterm elections, AI will become a scapegoat for the cost of living issue."

"The ROSA Act passed the House 300 to 20 but is stalled in the Senate. Corporate lobbying is blocking legislation that restricts China's remote access to GPUs."

Main Discussion

Best First Half in Semiconductor History, Then Paying the Piper

Dylan: We've got a market drawdown, all the AI names are down. Today, we're either going to pour fuel on the fire or offer some comfort.

Doug: Through June 30th, the end of Q2, it was probably the best performance in semiconductor history. Then we started to unwind. A lot of it can be attributed to technical factors: leverage, momentum reversal. But the reality is, the faster something rises, the harder gravity pulls it down. We're paying for the crazy momentum rally we had before.

The situation in Korea is crazy. Every day there are stocks hitting their daily limit down. There was a tweet saying "How do I do my job?" The HR director lost all their money, everyone is depressed because all stocks are down. If you look back at the history of Asian financial markets, this happens more often than you think.

One of my favorite books is about the great Taiwan bubble. Taiwan had a 100x bubble on a per capita basis, banks traded at 500 times earnings, everything was insane.

Dylan: When was this?

Doug: Late 1980s.

Dylan: Do you think the fundamentals in Korea now are different from back then?

Doug: Fundamentals are good. But the problem is, things are never as bad as you fear, nor as good as you imagine. SK Hynix missed expectations today because they shifted towards more LTAs. Ironically, during their ADR roadshow, they were criticizing Micron for doing LTAs at lower prices.

Memory prices roughly tripled last year. They can't triple again next year, maybe increase by 30 to 50%. But the financial world is completely broken, only looking at the rate of change. Historically, in memory cycles, once the second derivative comes down, it's usually the end. Because the rate of change doesn't stop at 30%; it goes straight to negative 50%.

The playbook for this cycle is always the same: everyone invests in factories, capacity comes online, and then they realize, "Oh my god, why is demand so low?" Because there was double ordering, triple ordering. Factory utilization drops from 100% to 50%, and the only way to recoup costs is to slash prices. That's the nature of the semiconductor market.

KOSPI is now down 40%. People with 2x leverage are wiped out. Then it's a self-fulfilling spiral: everyone watches their accounts shrink, decides to sell, which exacerbates the decline.

Chinese Memory: Could Spoil the Party, But Demand Still Exceeds Supply

Dylan: Recently, Chinese memory has entered the ecosystem. CXMT, YMTC are having big IPOs. What's your take?

Doug: Historically, whatever China touches, they turn it into a commodity. They have capacity, even if yields are low, they don’t care. Chinese companies aren't competing for profit margins or EPS; their shareholder is the government, which incentivizes production, and provinces compete with each other over GDP.

CXMT is clearly the fourth player in the market now, but in this shortage environment, they can still make money. Apple has started using CXMT's memory because Micron was "price gouging." Nobody is crying at the casino, Tim Apple. You have to buy at market prices.

CXMT might spoil the party, but the reality is demand still exceeds supply. The real trillion-dollar question is: Where is the demand? The supply curve is relatively easy to understand. We don't know the demand curve. We know coding agents and chatbots imply more demand, but whether it's 10x or 100x, we don't know. Supply will ramp up blindly until it hits the demand curve one day.

Coding Agents Are the Inflection Point: SemiAnalysis’s Own 100x AI Spend

Dylan: I think demand is clearly very strong and will remain so for a long time. I just look at our own company's internal usage. If you think future demand will flatten or decline, you have to believe the models won't get any better. I don't see any signs of stagnation, only signals pointing the other way.

Doug: Let me play devil's advocate. What is the biggest bear argument? The speed of technological progress might outpace the rate at which people use it. Let's say AI's killer app is data entry; a Kimi K3 would be sufficient. We build faster cars and better products, but the real demand curve is already met by a product we have.

It's like the internet bubble: they said "demand doubles every 90 days," but fiber optic technology improved 2 to 3 times every year. In the end, a single fiber's capacity became 500,000 times greater, and everyone said, "Wait, maybe we don't need this much fiber."

Dylan: I disagree, but it's worth discussing. My counter is: there are 100 to 1000 times more people who aren't using any models right now. Second, AI's use cases go far beyond coding. It can do video generation, drug discovery, material science. Someone is using AI for superconducting components. How much is that worth? It's worth a lot of GPUs.

And coding itself isn't just "centering a div." It represents a whole class of tasks with much higher economic value than front-end debugging. Sam Altman talks about RSI (Recursive Self-Improvement), Anthropic has new models coming out. Coding agents with Claude 4.5 were a clear inflection point: you cross a certain intelligence threshold, and a whole new market appears. What you couldn't do yesterday, you can do today.

Doug: You are the prototype user. Last year, less than 10 people on SemiAnalysis's tech team were using coding agents. Then you and Dylan said "everyone in the company needs to learn this." Now we have 90 users.

Dylan: From 9 to 90, a 10x increase. Then within 3 to 4 months, per-person usage also increased about 10x. The company's AI spending grew 100x. The question is, will this happen for every company? Maybe not at our intensity, but many companies have a lot of work that can be cut.

H100s Won't Become Scrap, But Models Are Getting Bigger

Doug: I think old chips will become worthless. Everyone says "H100 is an appreciating asset," but one day, running a model might require 100 H100s. Then you'll say "retire the old girl, buy a B300." The real confirmation signal will be pricing divergence between B200 and B300.

Dylan: I completely disagree. The most fundamental reason: No one will rip out H100s and replace them with B300s. Data center designs are completely different. You can't replace Hopper with Blackwell or Rubin in the same facility; you have to tear the whole thing down and rebuild. So to justify retiring an entire Hopper data center, you first have to prove the revenue from those chips is below the operating cost. This isn't a variable cost; it's a sunk cost.

Doug: In a frictionless world, you'd be right, but we live in a world with increasing friction. The friction for building new compute includes power permits, land, and approvals.

Dylan: Right, I agree. The scenario for GPU prices falling is if model progress stalls. The scenario for them rising is if model progress continues. There's another X factor: government intervention with frontier labs. If limits are placed on who can use the newest, best chips, demand gets compressed, and the price of older chips falls too.

Capital and Electricians: The Physical Ceilings of Scaling Laws

Doug: What worries me most isn't demand; it's the physical bottlenecks on the supply side. First are electricians. The US is short 100,000 electricians. A mid-level electrician makes $250k a year. Those willing to work 18-hour days can get $400k-$500k. There's a website tracking electrician job postings. On the Wayback Machine, you can see hourly rates go from $15-$20 to $50, $100, $200. It takes 18 months to train an electrician. We've never trained enough to double that number again.

Second is capital. Hyperscalers have issued about $450 billion in debt this year, the largest amount ever. Second only to the US government and the Chinese government. Someone has to buy this debt. To get them to buy more, you have to offer higher interest rates. And the source of this money is largely pensions and annuities. The pension pool is structurally shrinking. Pensions have largely moved to 401ks, and 401ks don't buy debt. So you basically have to believe everyone needs twice the insurance, which doesn't make sense.

Scaling laws say "Great, let's make the model twice as big." But not everything can scale up twofold or threefold synchronously.

Dylan: Wait, are you saying pensions are paying for data center construction?

Doug: Yes. Annuities are bought right before retirement. The baby boomers are all retiring, so this asset pool is relatively large. But can it double? Can it triple? I don't think so. Life insurance is another source. But you have to believe everyone needs twice the insurance. No one is going to buy twice the life insurance.

Dylan: This is fascinating. Pensions are structurally shrinking, but there is indeed a lot of money there.

Doug: Another example is Taiwan. TSMC directly and indirectly accounts for 20% of Taiwan's GDP. If TSMC doubles or triples again, Taiwan would need to have more babies just to get enough workers. Taiwan only has one game in play. Taiwan's GDP grew 25% this year, just from TSMC baking chips. But to double again, there won't be enough people.

ลงทุน

เทคโนโลยี

AI