AI Peak Debate: What Big Money and Manufacturing Data Reveal About the Next Market Phase
Part 1. In the AI Peak Debate, What Are the “Big Money” Investors Really Looking At?
From 2024 through 2025, global financial markets have basically revolved around one topic: AI.
NVIDIA has dominated the GPU market, Meta has been scaling up AI models and data centers, and Google and Microsoft have been pouring tens of billions of dollars into the generative AI race to secure the lead.
This naturally shaped the overall direction of the equity markets.
In fact, a large chunk of the S&P 500’s gains between 2023 and 2025 came from the “Magnificent 7 (M7)” mega-cap tech stocks.
Some analysts even went as far as to say:
That’s how dominant the theme has been.
But in the second half of 2025, the tone started to change.
The market began to ask a tougher question:
To answer that, many investors naturally turned their eyes to the “big money” — the world-class investors who move billions at a time.
The logic is simple.
The cleanest way to track those moves is through the 13F filings the U.S. Securities and Exchange Commission (SEC) publishes each quarter.
Any institution managing over $100 million has to disclose what it owns and in what size, so these filings have become known as:
1. Warren Buffett — Why Did He Finally Buy Google Right Before Retiring?
The most eye-catching detail in the 2025 Q3 13F filings was the shift inside Warren Buffett’s portfolio.
What Berkshire Hathaway did in Q3 2025
- Initiated a new position in Alphabet (Google): 17,846,142 shares
- Fair value of roughly $4.3 billion (about 6.3 trillion KRW)
- Google instantly became Berkshire’s 10th largest holding (1.6% weight)
- To fund the purchase, Berkshire sold about 20% of its Apple stake — Apple shares sold worth roughly $10.6 billion
Buffett has stuck to his value investing philosophy for over 60 years and has always been very cautious with big tech.
The last time he initiated a new mega-cap tech position was Amazon back in 2019.
Yet on the verge of retirement in 2025, he chose Google.
So why Google?
“Google is the stock I regret the most not buying. It’s not easy to buy something that’s already gone up ten-fold.”
Since that comment, Google went on to climb another 480% or so, becoming a textbook example of winner-takes-all.
But the real takeaway here is not just “Buffett bought Google.”
Google has:
- A powerful cash-engine in search and advertising
- Growing cloud revenue, especially on the back of AI demand
- YouTube’s overwhelming influence as a global platform
In other words, it’s not just a “tech story” — it’s a business with deeply established fundamentals.
Once news broke that Buffett had bought Google, the stock jumped about 4% in after-hours trading.
His move clearly signaled more than just another portfolio tweak.
2. Stanley Druckenmiller — “AI Is a Huge Opportunity, but Not Everyone Will Win”
On the other side of the spectrum, legendary trader Stanley Druckenmiller made a very different set of choices.
Druckenmiller’s key Q3 moves
- Bought roughly 430,000 shares of Amazon
- Bought about 70,000 shares of Meta
- Sold his entire 200,000-share position in Microsoft
That mix is pretty intriguing.
Amazon and Meta are both ramping up AI investments, while at the same time seeing a renewed upswing in their core businesses — e-commerce, digital ads, and cloud.
Microsoft, on the other hand, sits right at the front line of AI through its partnership with OpenAI, but many analysts have argued the stock moved into expensive territory faster than its fundamentals.
At the same time, Druckenmiller ramped up his exposure to health care names:
- Natera
- Insmed
- Teva
All three share a common trait: they’re “defensive growth” names with tangible earnings backing their stories.
His own words probably summarize it best:
3. Bridgewater and Tiger Global — Starting to Dial Back AI-Heavy Portfolios
Some of the hedge funds that rode the AI bull wave most aggressively also started trimming their positions in 2025.
Bridgewater Associates
As the world’s largest hedge fund, Bridgewater’s shifts often reflect broader market sentiment.
- Cut its NVIDIA stake by roughly two-thirds
- Sold more than half its Google position
This looks less like routine rebalancing and more like a deliberate reduction in AI exposure amid peak-cycle concerns.
Tiger Global and Lone Pine Capital
- Meta positions:
- Tiger Global: reduced by 62.6%
- Lone Pine: reduced by 34.8%
AI-driven platform companies have seen their earnings ramp strongly, but their stock prices have climbed even faster, pushing valuations into much more demanding territory.
Why is this happening now?
Expanding AI infrastructure requires huge upfront spending, regardless of near-term profits:
- Building or leasing large-scale data centers
- Buying GPUs in bulk
- Higher model training and inference costs
All of that tends to come with growing leverage and larger balance sheets.
So underneath the strong earnings, there’s a competing narrative:
That’s why some hedge funds have been happy to lock in gains and shift into a more risk-controlled stance.
Summary — What the Big Money Is Really Saying
We can sum up Part 1 like this:
Buffett
- “Proven tech franchises are still attractive at the right price.”
- Big new position in Google, partial trim in Apple
Druckenmiller
- “Be selective in AI; health care is a solid alternative.”
- Added to Amazon and Meta, exited Microsoft
Bridgewater and Tiger Global
- “We’re worried about overheating in AI — time to de-risk.”
- Reduced stakes in NVIDIA, Meta, and Google
In short, even among top professionals, there’s no single consensus on how fast to lean into AI or how much risk is acceptable at this stage.
Part 2. The Risks You Don’t See on the Surface
The AI industry is still at the center of global growth. But outside the spotlight, in areas most retail investors never look at, there’s a set of “shadow indicators” that have been flashing more brightly.
On the surface, the narrative sounds like this:
That’s the positive story everyone knows.
But deep in the plumbing of the financial system, very different data points are starting to show up.
The first places to react are the derivatives markets and the cash-flow metrics and growth rates of the mega-cap tech names. In Part 2, we’ll walk through some of the warning signs most people don’t talk about.
4. What the Derivatives Market Is Telling Us: Default Risk Is Creeping Up in Big Tech
The AI ecosystem is an incredibly capital-intensive business. Expanding data centers, buying GPUs by the truckload, and securing power infrastructure all require tens of billions of dollars in CAPEX.
Most of that funding comes from:
- Issuing corporate bonds
- Bank loans
- Long-term debt on the balance sheet
So alongside AI-driven growth, there’s a parallel trend we can’t ignore:
Big Tech’s debt loads are rising fast.
Bloomberg’s focus: a spike in CDS activity
Bloomberg recently pointed out that as AI investment has accelerated, trading in credit default swaps (CDS) on major tech names has surged at one of the fastest paces on record.
CDS volume is one of the cleanest ways to see how actively the market is hedging credit risk.
5. Oracle’s CDS Volume — A Small Snapshot of a Bigger Shift
| Period | CDS Notional Volume |
|---|---|
| Last year | About $200 million |
| Same period this year | Roughly $4.2 billion |
That’s a 21x jump in just a year.
Oracle has been leaning into AI infrastructure — more data centers, more GPUs, more cloud capacity — and has been financing that with substantial new debt issuance.
The surge in CDS trading is the market’s way of saying:
6. Meta’s CDS Activity Also Picking Up
Meta recently issued around $30 billion in new bonds (roughly 40 trillion KRW).
Not long after, trading in CDS tied to Meta debt picked up sharply.
CDS, at its core, is default insurance. If the company goes into default, the CDS can pay out like an insurance policy. So when you see a lot of CDS changing hands, it usually means:
7. Why This Matters More Than It Looks
On the surface, it still looks like a feel-good story:
- Revenue is growing
- AI investment is ramping up
- Innovation headlines keep coming
So it’s easy to fall into a purely bullish narrative.
During the 2008 financial crisis, the earliest signs of trouble showed up in CDS quotes long before most stock investors realized what was happening. The same pattern — more hedging, rising concern about leverage — is appearing again in the AI era.
8. The Biggest Swing Factor in the Whole Story: NVIDIA’s Slowing Growth
At the center of the AI boom is NVIDIA. It holds an 80–90% share of the high-end AI GPU market, and just about every serious AI workload depends on its chips.
But while revenues have kept growing, NVIDIA’s growth rate itself has been quietly decelerating.
9. How NVIDIA’s Growth Profile Is Changing
| Period | Revenue Growth |
|---|---|
| Past peak phase | Over +100% YoY |
| Recent (Q2 2025) | +54% YoY |
| Next 12–18 months (consensus) | +20–30% expected |
The company is still growing quickly in absolute terms, but the key point is this:
The growth rate has already been cut by more than half.
For a high-multiple growth stock, that’s not a trivial shift. For a name like NVIDIA, where the P/E ratio has been very rich, any visible slowdown in growth can lead to an outsized reaction in the stock price.
10. What One CIO Is Warning About
Yoon Je-sung, CIO at New York Life Investment Management, put it this way:
NVIDIA has posted revenue growth for six consecutive quarters, but the market is no longer focused only on whether it’s growing. The real question now is:
“At what pace can that growth realistically be sustained?”
We’ve shifted from a world where “growth exists” was enough, to a world where “the slope of that growth curve” is what really moves markets.
11. Michael Burry — The Bubble Cassandra Returns
Michael Burry — the real-life protagonist of The Big Short — has once again stepped in to warn the market in 2024–2025.
This time, he bought a large amount of put options on NVIDIA and Palantir, effectively betting on a decline.
The market’s response was immediate:
- Short-term pullbacks in NVIDIA
- A noticeable drop in Palantir’s stock
- Public pushback from some CEOs
Palantir CEO Alex Karp even went on record saying:
Burry, however, didn’t back down from his thesis.
When he later announced he was winding down his fund and stepping away from managing outside money, he left the market with a very symbolic final message:
In other words, the real risk isn’t AI as a technology — it’s the over-excitement and leverage building up around it.
12. Part 2 Takeaways
AI is still reshaping industries and creating real economic value. But beneath that story, we’re also seeing:
- Rising debt levels at Big Tech
- A surge in CDS trading as more investors hedge credit risk
- A visible slowdown in NVIDIA’s growth rate
- High-profile warnings from investors like Michael Burry
Those signals don’t automatically mean the AI bubble bursts tomorrow. But they do mean that any serious investment strategy now has to account for slowing growth on the earnings side and increasing leverage on the balance sheet.
Part 3. What Manufacturing Data Is Telling Us
From the outside, the AI era looks like a tech-driven boom — hyper-growth in semis, cloud, and software. But deep in the macro data, something else is quietly steering the ultimate direction of risk assets.
That something is the real economy, and the most important piece of it is still manufacturing.
Yes, AI, semiconductors, and cloud companies are putting up impressive numbers, and it’s easy to argue “the AI bubble hasn’t even started.” But many institutions and strategists are now saying:
If you want to know whether we’re near the top of the AI cycle, don’t just look at tech stocks — look at manufacturing PMIs.
The reasoning is straightforward:
13. U.S. Manufacturing PMI — 52.5 and Still Expanding, but With Softer Export Signals
Let’s start with the U.S., the world’s largest consumer market and the anchor for many global investment decisions.
According to S&P Global, the latest U.S. Manufacturing PMI came in at 52.5. A reading above 50 means the sector is in expansion versus the previous month.
So at a headline level, we can say:
U.S. manufacturing is still relatively solid.
But once you dig into the sub-components, the picture gets more nuanced:
- New export orders are slowing
- Business expectations for the future are slipping
- Inventories are building up
- The pace of new orders for raw materials is moderating
All of that adds up to a message along the lines of “U.S. manufacturing is expanding, but not at the same speed as before.”
And that matters, because weaker export momentum in the U.S. quickly spills over to export-driven economies like Korea, Taiwan, and Japan.
14. Korea’s Manufacturing PMI — Back Below 50 at 49.4
Korea’s latest Manufacturing PMI reading is 49.4. Just a month earlier it had finally climbed back up to 50.7, marking an expansion after roughly eight months of contraction.
But within a month, it slipped back below 50.
Because Korea’s PMI is heavily influenced by exports, the mixed order trends in key sectors such as:
- Semiconductors
- Autos
- Displays
- EV batteries and related components
have all been weighing on the headline number.
On top of that, Korea’s average manufacturing capacity utilization has been stuck around the 71–73% range. Many economists would say you need something closer to the 80% level before you can call it a robust, self-sustaining expansion.
In other words, Korea is still hovering somewhere between a soft patch and a full-on recovery.
15. Why Manufacturing Matters So Much — The Main Lesson from the Dot-Com Bubble
If you go back to the 1995–2000 dot-com period, the real tipping point wasn’t just that tech stocks were expensive.
The deeper issue was that the real economy started to lose momentum underneath that tech optimism.
DB Financial Investment’s analysis of that era highlights a simple but powerful conclusion:
Back then, many internet business models were still unproven. Once you layered in higher interest rates, weaker manufacturing data, and softer exports, investor confidence began to erode — and that erosion is what ultimately popped the bubble.
The AI cycle has clear parallels:
- Huge technological potential
- Explosive investment flows
- Rapid stock price appreciation
- Large-scale capex cycles
- Rising expectations — and then over-expectations
There is one important difference, though. AI is translating into real earnings much faster than most dot-com names ever did. Still, if the underlying real economy falters, trust in high-growth tech will be the first thing to wobble.
So where are we now?
- U.S. manufacturing: holding up, but showing softer edges
- Korea: brief stabilization, then back to mild contraction
- Europe: stuck in a longer-term slowdown
- China: oscillating between recovery and renewed weakness
So the simple summary is: AI looks strong, but manufacturing doesn’t. The bigger that disconnect gets, the more careful markets become — and the more investors are forced to be selective.
16. Bottom Line — In the AI Era, Expect a “Barbell” Style of Investing
By late 2025, two distinct investment patterns are already co-existing in the global market.
The Buffett Playbook
“Keep buying high-quality tech names with proven earnings.”
Buffett’s decision to initiate a big new position in Google sends a clear message:
Names like Google, Amazon, and Meta were already cash machines even before the AI wave hit. For them, AI is more of an earnings accelerator than a speculative side bet.
The Druckenmiller and Bridgewater Playbook
“Dial down AI exposure at the margins and rotate into defenses.”
This camp doesn’t deny the long-term potential of AI, but it is very sensitive to near-term overheating, higher leverage, and slowing growth rates.
That’s why they’ve been trimming positions in NVIDIA, Meta, and Google, and adding to health care, staples, and undervalued value stocks.
17. What Both Sides Actually Agree On
These two strategies look very different on the surface, but they share a core belief.
So a sensible stance toward AI right now is neither blind optimism nor full-blown pessimism.
The key is to use real-economy data — especially manufacturing and credit — to decide which tech names actually deserve your capital.
18. Three Macro Variables That Could Decide 2025’s Market Path
Heading into the end of 2025, three variables are likely to play an outsized role in where global equities go from here:
- The pace of NVIDIA’s growth deceleration
- The strength of U.S. manufacturing (especially PMI and exports)
- The cost of capital for Big Tech (rising debt and interest expenses)
These aren’t just numbers on a page. Together, they’re a stress test of the AI era’s “base layer” — the underlying economic and financial muscle that has to support all that growth.
References
U.S. Equity Market and 13F Data
- U.S. Securities and Exchange Commission (SEC) — Form 13F filings
- Bloomberg — Reports on rising Big Tech CDS volumes
- Reuters and CNBC — Coverage of NVIDIA, Meta, Amazon and other “Magnificent 7” portfolio shifts
Source Articles (Summarized / Referenced)
- Major Korean economic dailies (Korea Economic Daily, Maeil Business, Seoul Economic Daily)
- Articles on Warren Buffett’s new position in Alphabet (Google)
- Reports on Stanley Druckenmiller’s Q3 portfolio changes
- Coverage of Bridgewater and Tiger Global trimming AI exposure
- Articles on surging CDS activity in Big Tech
Global Manufacturing Data
- S&P Global Manufacturing PMI — U.S. manufacturing PMI (52.5)
- TradingEconomics — Korea’s manufacturing PMI (49.4) and other major economies’ PMI data
- Ministry of Economy and Finance and Korea Institute for Industrial Economics & Trade — Data on Korean manufacturing capacity utilization
AI and Semiconductor Market Analysis
- Bloomberg Intelligence — NVIDIA growth outlook and AI-related capex analysis
- IMF and BIS publications — Discussions of bubbles, liquidity cycles, and new-economy investment booms
Part 1. In the AI Peak Debate, What Are the “Big Money” Investors Really Looking At?
From 2024 through 2025, global financial markets have basically revolved around one topic: AI.
NVIDIA has dominated the GPU market, Meta has been scaling up AI models and data centers, and Google and Microsoft have been pouring tens of billions of dollars into the generative AI race to secure the lead.
This naturally shaped the overall direction of the equity markets.
In fact, a large chunk of the S&P 500’s gains between 2023 and 2025 came from the “Magnificent 7 (M7)” mega-cap tech stocks.
Some analysts even went as far as to say:
That’s how dominant the theme has been.
But in the second half of 2025, the tone started to change.
The market began to ask a tougher question:
To answer that, many investors naturally turned their eyes to the “big money” — the world-class investors who move billions at a time.
The logic is simple.
The cleanest way to track those moves is through the 13F filings the U.S. Securities and Exchange Commission (SEC) publishes each quarter.
Any institution managing over $100 million has to disclose what it owns and in what size, so these filings have become known as:
1. Warren Buffett — Why Did He Finally Buy Google Right Before Retiring?
The most eye-catching detail in the 2025 Q3 13F filings was the shift inside Warren Buffett’s portfolio.
What Berkshire Hathaway did in Q3 2025
- Initiated a new position in Alphabet (Google): 17,846,142 shares
- Fair value of roughly $4.3 billion (about 6.3 trillion KRW)
- Google instantly became Berkshire’s 10th largest holding (1.6% weight)
- To fund the purchase, Berkshire sold about 20% of its Apple stake — Apple shares sold worth roughly $10.6 billion
Buffett has stuck to his value investing philosophy for over 60 years and has always been very cautious with big tech.
The last time he initiated a new mega-cap tech position was Amazon back in 2019.
Yet on the verge of retirement in 2025, he chose Google.
So why Google?
“Google is the stock I regret the most not buying. It’s not easy to buy something that’s already gone up ten-fold.”
Since that comment, Google went on to climb another 480% or so, becoming a textbook example of winner-takes-all.
But the real takeaway here is not just “Buffett bought Google.”
Google has:
- A powerful cash-engine in search and advertising
- Growing cloud revenue, especially on the back of AI demand
- YouTube’s overwhelming influence as a global platform
In other words, it’s not just a “tech story” — it’s a business with deeply established fundamentals.
Once news broke that Buffett had bought Google, the stock jumped about 4% in after-hours trading.
His move clearly signaled more than just another portfolio tweak.
2. Stanley Druckenmiller — “AI Is a Huge Opportunity, but Not Everyone Will Win”
On the other side of the spectrum, legendary trader Stanley Druckenmiller made a very different set of choices.
Druckenmiller’s key Q3 moves
- Bought roughly 430,000 shares of Amazon
- Bought about 70,000 shares of Meta
- Sold his entire 200,000-share position in Microsoft
That mix is pretty intriguing.
Amazon and Meta are both ramping up AI investments, while at the same time seeing a renewed upswing in their core businesses — e-commerce, digital ads, and cloud.
Microsoft, on the other hand, sits right at the front line of AI through its partnership with OpenAI, but many analysts have argued the stock moved into expensive territory faster than its fundamentals.
At the same time, Druckenmiller ramped up his exposure to health care names:
- Natera
- Insmed
- Teva
All three share a common trait: they’re “defensive growth” names with tangible earnings backing their stories.
His own words probably summarize it best:
3. Bridgewater and Tiger Global — Starting to Dial Back AI-Heavy Portfolios
Some of the hedge funds that rode the AI bull wave most aggressively also started trimming their positions in 2025.
Bridgewater Associates
As the world’s largest hedge fund, Bridgewater’s shifts often reflect broader market sentiment.
- Cut its NVIDIA stake by roughly two-thirds
- Sold more than half its Google position
This looks less like routine rebalancing and more like a deliberate reduction in AI exposure amid peak-cycle concerns.
Tiger Global and Lone Pine Capital
- Meta positions:
- Tiger Global: reduced by 62.6%
- Lone Pine: reduced by 34.8%
AI-driven platform companies have seen their earnings ramp strongly, but their stock prices have climbed even faster, pushing valuations into much more demanding territory.
Why is this happening now?
Expanding AI infrastructure requires huge upfront spending, regardless of near-term profits:
- Building or leasing large-scale data centers
- Buying GPUs in bulk
- Higher model training and inference costs
All of that tends to come with growing leverage and larger balance sheets.
So underneath the strong earnings, there’s a competing narrative:
That’s why some hedge funds have been happy to lock in gains and shift into a more risk-controlled stance.
Summary — What the Big Money Is Really Saying
We can sum up Part 1 like this:
Buffett
- “Proven tech franchises are still attractive at the right price.”
- Big new position in Google, partial trim in Apple
Druckenmiller
- “Be selective in AI; health care is a solid alternative.”
- Added to Amazon and Meta, exited Microsoft
Bridgewater and Tiger Global
- “We’re worried about overheating in AI — time to de-risk.”
- Reduced stakes in NVIDIA, Meta, and Google
In short, even among top professionals, there’s no single consensus on how fast to lean into AI or how much risk is acceptable at this stage.
Part 2. The Risks You Don’t See on the Surface
The AI industry is still at the center of global growth. But outside the spotlight, in areas most retail investors never look at, there’s a set of “shadow indicators” that have been flashing more brightly.
On the surface, the narrative sounds like this:
That’s the positive story everyone knows.
But deep in the plumbing of the financial system, very different data points are starting to show up.
The first places to react are the derivatives markets and the cash-flow metrics and growth rates of the mega-cap tech names. In Part 2, we’ll walk through some of the warning signs most people don’t talk about.
4. What the Derivatives Market Is Telling Us: Default Risk Is Creeping Up in Big Tech
The AI ecosystem is an incredibly capital-intensive business. Expanding data centers, buying GPUs by the truckload, and securing power infrastructure all require tens of billions of dollars in CAPEX.
Most of that funding comes from:
- Issuing corporate bonds
- Bank loans
- Long-term debt on the balance sheet
So alongside AI-driven growth, there’s a parallel trend we can’t ignore:
Big Tech’s debt loads are rising fast.
Bloomberg’s focus: a spike in CDS activity
Bloomberg recently pointed out that as AI investment has accelerated, trading in credit default swaps (CDS) on major tech names has surged at one of the fastest paces on record.
CDS volume is one of the cleanest ways to see how actively the market is hedging credit risk.
5. Oracle’s CDS Volume — A Small Snapshot of a Bigger Shift
| Period | CDS Notional Volume |
|---|---|
| Last year | About $200 million |
| Same period this year | Roughly $4.2 billion |
That’s a 21x jump in just a year.
Oracle has been leaning into AI infrastructure — more data centers, more GPUs, more cloud capacity — and has been financing that with substantial new debt issuance.
The surge in CDS trading is the market’s way of saying:
6. Meta’s CDS Activity Also Picking Up
Meta recently issued around $30 billion in new bonds (roughly 40 trillion KRW).
Not long after, trading in CDS tied to Meta debt picked up sharply.
CDS, at its core, is default insurance. If the company goes into default, the CDS can pay out like an insurance policy. So when you see a lot of CDS changing hands, it usually means:
7. Why This Matters More Than It Looks
On the surface, it still looks like a feel-good story:
- Revenue is growing
- AI investment is ramping up
- Innovation headlines keep coming
So it’s easy to fall into a purely bullish narrative.
During the 2008 financial crisis, the earliest signs of trouble showed up in CDS quotes long before most stock investors realized what was happening. The same pattern — more hedging, rising concern about leverage — is appearing again in the AI era.
8. The Biggest Swing Factor in the Whole Story: NVIDIA’s Slowing Growth
At the center of the AI boom is NVIDIA. It holds an 80–90% share of the high-end AI GPU market, and just about every serious AI workload depends on its chips.
But while revenues have kept growing, NVIDIA’s growth rate itself has been quietly decelerating.
9. How NVIDIA’s Growth Profile Is Changing
| Period | Revenue Growth |
|---|---|
| Past peak phase | Over +100% YoY |
| Recent (Q2 2025) | +54% YoY |
| Next 12–18 months (consensus) | +20–30% expected |
The company is still growing quickly in absolute terms, but the key point is this:
The growth rate has already been cut by more than half.
For a high-multiple growth stock, that’s not a trivial shift. For a name like NVIDIA, where the P/E ratio has been very rich, any visible slowdown in growth can lead to an outsized reaction in the stock price.
10. What One CIO Is Warning About
Yoon Je-sung, CIO at New York Life Investment Management, put it this way:
NVIDIA has posted revenue growth for six consecutive quarters, but the market is no longer focused only on whether it’s growing. The real question now is:
“At what pace can that growth realistically be sustained?”
We’ve shifted from a world where “growth exists” was enough, to a world where “the slope of that growth curve” is what really moves markets.
11. Michael Burry — The Bubble Cassandra Returns
Michael Burry — the real-life protagonist of The Big Short — has once again stepped in to warn the market in 2024–2025.
This time, he bought a large amount of put options on NVIDIA and Palantir, effectively betting on a decline.
The market’s response was immediate:
- Short-term pullbacks in NVIDIA
- A noticeable drop in Palantir’s stock
- Public pushback from some CEOs
Palantir CEO Alex Karp even went on record saying:
Burry, however, didn’t back down from his thesis.
When he later announced he was winding down his fund and stepping away from managing outside money, he left the market with a very symbolic final message:
In other words, the real risk isn’t AI as a technology — it’s the over-excitement and leverage building up around it.
12. Part 2 Takeaways
AI is still reshaping industries and creating real economic value. But beneath that story, we’re also seeing:
- Rising debt levels at Big Tech
- A surge in CDS trading as more investors hedge credit risk
- A visible slowdown in NVIDIA’s growth rate
- High-profile warnings from investors like Michael Burry
Those signals don’t automatically mean the AI bubble bursts tomorrow. But they do mean that any serious investment strategy now has to account for slowing growth on the earnings side and increasing leverage on the balance sheet.
Part 3. What Manufacturing Data Is Telling Us
From the outside, the AI era looks like a tech-driven boom — hyper-growth in semis, cloud, and software. But deep in the macro data, something else is quietly steering the ultimate direction of risk assets.
That something is the real economy, and the most important piece of it is still manufacturing.
Yes, AI, semiconductors, and cloud companies are putting up impressive numbers, and it’s easy to argue “the AI bubble hasn’t even started.” But many institutions and strategists are now saying:
If you want to know whether we’re near the top of the AI cycle, don’t just look at tech stocks — look at manufacturing PMIs.
The reasoning is straightforward:
13. U.S. Manufacturing PMI — 52.5 and Still Expanding, but With Softer Export Signals
Let’s start with the U.S., the world’s largest consumer market and the anchor for many global investment decisions.
According to S&P Global, the latest U.S. Manufacturing PMI came in at 52.5. A reading above 50 means the sector is in expansion versus the previous month.
So at a headline level, we can say:
U.S. manufacturing is still relatively solid.
But once you dig into the sub-components, the picture gets more nuanced:
- New export orders are slowing
- Business expectations for the future are slipping
- Inventories are building up
- The pace of new orders for raw materials is moderating
All of that adds up to a message along the lines of “U.S. manufacturing is expanding, but not at the same speed as before.”
And that matters, because weaker export momentum in the U.S. quickly spills over to export-driven economies like Korea, Taiwan, and Japan.
14. Korea’s Manufacturing PMI — Back Below 50 at 49.4
Korea’s latest Manufacturing PMI reading is 49.4. Just a month earlier it had finally climbed back up to 50.7, marking an expansion after roughly eight months of contraction.
But within a month, it slipped back below 50.
Because Korea’s PMI is heavily influenced by exports, the mixed order trends in key sectors such as:
- Semiconductors
- Autos
- Displays
- EV batteries and related components
have all been weighing on the headline number.
On top of that, Korea’s average manufacturing capacity utilization has been stuck around the 71–73% range. Many economists would say you need something closer to the 80% level before you can call it a robust, self-sustaining expansion.
In other words, Korea is still hovering somewhere between a soft patch and a full-on recovery.
15. Why Manufacturing Matters So Much — The Main Lesson from the Dot-Com Bubble
If you go back to the 1995–2000 dot-com period, the real tipping point wasn’t just that tech stocks were expensive.
The deeper issue was that the real economy started to lose momentum underneath that tech optimism.
DB Financial Investment’s analysis of that era highlights a simple but powerful conclusion:
Back then, many internet business models were still unproven. Once you layered in higher interest rates, weaker manufacturing data, and softer exports, investor confidence began to erode — and that erosion is what ultimately popped the bubble.
The AI cycle has clear parallels:
- Huge technological potential
- Explosive investment flows
- Rapid stock price appreciation
- Large-scale capex cycles
- Rising expectations — and then over-expectations
There is one important difference, though. AI is translating into real earnings much faster than most dot-com names ever did. Still, if the underlying real economy falters, trust in high-growth tech will be the first thing to wobble.
So where are we now?
- U.S. manufacturing: holding up, but showing softer edges
- Korea: brief stabilization, then back to mild contraction
- Europe: stuck in a longer-term slowdown
- China: oscillating between recovery and renewed weakness
So the simple summary is: AI looks strong, but manufacturing doesn’t. The bigger that disconnect gets, the more careful markets become — and the more investors are forced to be selective.
16. Bottom Line — In the AI Era, Expect a “Barbell” Style of Investing
By late 2025, two distinct investment patterns are already co-existing in the global market.
The Buffett Playbook
“Keep buying high-quality tech names with proven earnings.”
Buffett’s decision to initiate a big new position in Google sends a clear message:
Names like Google, Amazon, and Meta were already cash machines even before the AI wave hit. For them, AI is more of an earnings accelerator than a speculative side bet.
The Druckenmiller and Bridgewater Playbook
“Dial down AI exposure at the margins and rotate into defenses.”
This camp doesn’t deny the long-term potential of AI, but it is very sensitive to near-term overheating, higher leverage, and slowing growth rates.
That’s why they’ve been trimming positions in NVIDIA, Meta, and Google, and adding to health care, staples, and undervalued value stocks.
17. What Both Sides Actually Agree On
These two strategies look very different on the surface, but they share a core belief.
So a sensible stance toward AI right now is neither blind optimism nor full-blown pessimism.
The key is to use real-economy data — especially manufacturing and credit — to decide which tech names actually deserve your capital.
18. Three Macro Variables That Could Decide 2025’s Market Path
Heading into the end of 2025, three variables are likely to play an outsized role in where global equities go from here:
- The pace of NVIDIA’s growth deceleration
- The strength of U.S. manufacturing (especially PMI and exports)
- The cost of capital for Big Tech (rising debt and interest expenses)
These aren’t just numbers on a page. Together, they’re a stress test of the AI era’s “base layer” — the underlying economic and financial muscle that has to support all that growth.
References
U.S. Equity Market and 13F Data
- U.S. Securities and Exchange Commission (SEC) — Form 13F filings
- Bloomberg — Reports on rising Big Tech CDS volumes
- Reuters and CNBC — Coverage of NVIDIA, Meta, Amazon and other “Magnificent 7” portfolio shifts
Source Articles (Summarized / Referenced)
- Major Korean economic dailies (Korea Economic Daily, Maeil Business, Seoul Economic Daily)
- Articles on Warren Buffett’s new position in Alphabet (Google)
- Reports on Stanley Druckenmiller’s Q3 portfolio changes
- Coverage of Bridgewater and Tiger Global trimming AI exposure
- Articles on surging CDS activity in Big Tech
Global Manufacturing Data
- S&P Global Manufacturing PMI — U.S. manufacturing PMI (52.5)
- TradingEconomics — Korea’s manufacturing PMI (49.4) and other major economies’ PMI data
- Ministry of Economy and Finance and Korea Institute for Industrial Economics & Trade — Data on Korean manufacturing capacity utilization
AI and Semiconductor Market Analysis
- Bloomberg Intelligence — NVIDIA growth outlook and AI-related capex analysis
- IMF and BIS publications — Discussions of bubbles, liquidity cycles, and new-economy investment booms
댓글
댓글 쓰기