0:18大家好。今天是就业数据周五，又赶上美国的劳动节长周末，先祝大家劳动节快乐。这个月的 In the Know 会有点不一样：我们要看一批时间跨度非常长的图表，因为我们在试着把工业革命和这场技术革命联系起来。这件事我们还在做，所以我提出的问题可能和给出的答案一样多。我们一起走这一趟，把拼图一块块拼起来，尤其是通胀和利率，特别是现在我们对美联储主席 Warsh 的想法了解得多得多了。他在 Jackson Hole 的讲话很有启发，这个我们也会讲到。
6:02我为什么要讲这些？因为它和通胀、利率关系重大。名义 GDP 增速等于实际 GDP 增速加上通胀。1971 年我们放弃金汇兑本位之后，你能看到名义增速冲进了两位数；那时利率和通胀最后都到了两位数。此后是长达 40 年的长期下行：名义 GDP 增速和利率一起下降，所以我们这行大多数人，直到最近都只见过利率往下走。新冠期间是最后一次利率大幅下降，此后我们一直看到利率上升。
7:48我们一直说，通胀会低于预期，甚至可能转负，我会用图表说明为什么。但站在另一边起作用的是实际 GDP 增速，我们认为它会显著加速。所以名义 GDP 里会有一场拔河：实际增长往上拉，通胀往下拉，也许拉到负数。
8:33从趋势上看，名义 GDP 增速和 10 年期美债收益率的相关性相当高。10 年期收益率从 2023 年起一直在一个区间里筑底，今年已经是第三年了；而名义 GDP 增速按 10 年移动平均看，似乎正在突破，人们会预期美债收益率跟上。这个移动平均对终点比较敏感。如果我们说对了，实际增长往一个方向走、通胀往另一个方向走，利率可能继续在这个水平附近筑底。如果实际 GDP 增速超过 7%，到了 Elon Musk 说的那个区间，而通胀只是略微为负，我们就会看到 10 年期美债收益率上行。这只是市场在运作。
47:36我们一直在仔细想比特币和黄金，以及两者都能提供的保险作用，因为我们认为这场技术革命会让很多公司陷入险境；如果出事的公司够多，就会有对手方风险。1800 年代有 200 家铁路公司破产，因为追逐铁路机会的资本太多了。我们不认为 AI 会这样。铁路是建在希望和祈祷上的，指望新世界会出现、收入最终会来。而 AI 的收入在尖叫，这个领域今天的投入资本回报率就很高。
The most bullish macro voice around casts AI as a deflationary boom that replays the Industrial Revolution, and takes on the overbuild-then-bust bears directly.
Part 1 of 6 · 0:18
AI is a deflationary boom replaying the Industrial Revolution
Global real growth has run about 3% for 125 years. Five innovation platforms will at least double it to 6%, and could even multiply it fivefold to 15% as the Industrial Revolution did.
AI is a deflationary boom replaying the Industrial Revolution
Global real growth has run about 3% for 125 years. Five innovation platforms will at least double it to 6%, and could even multiply it fivefold to 15% as the Industrial Revolution did. Read this part →
Rates are normalizing; the yield curve may stay inverted
Real growth lifts short rates while falling inflation holds down long ones. During the Industrial Revolution, before the Fed, the curve was inverted more than half the time. Read this part →
The debt is bearable; money growth isn't inflationary
The deficit fell unexpectedly; government debt to corporate equity is near a record low; M2 is up about 5% a year, and velocity is flattening with labor participation. Read this part →
Inflation will fall hard: oil has peaked, technology collapses costs
Headline PCE at 3.7% is a temporary oil effect from the Iran war. Abu Dhabi left OPEC and raised output 78%; oil could return to $30; sequencing and inference costs are collapsing. Read this part →
AI creates jobs, and capital spending has barely begun
August payrolls rose 162,000; companies using AI more aggressively hire faster; non-defense capital goods broke out of a 20-plus-year range after ChatGPT. Read this part →
AI isn't the railroads: revenue is screaming, credit is calm
200 railroads went bankrupt in the 1800s; AI is different: Anthropic pays $50B per GW. There will be creative destruction, but credit default swaps and spreads are calm. Read this part →
Indigo's conclusion
She is the extreme bullish pole and is best read as a named counterparty. The evidence she uses against the bears is the very structure the bears call fragile: Anthropic's $50B per GW is a forward commitment, not a realized return.
How to read this The speaker is ARK Invest's founder and CIO, and ARK's whole thesis is a deflationary technology boom, high real growth and buying disruptive innovation. Every dial in this episode is turned to what suits ARK best (growth up fivefold to 15%, oil back to $30, inflation turning negative). Read it as an extreme bullish framework, not a forecast; she has made several high-profile deflationary-boom calls that didn't pan out.
What to remember
She shares Konstantine's Industrial Revolution analogy, and both step around something: Konstantine the speed of transition, Cathie the fragility of financing.
Her deflation case rests on two pillars: peak oil and technology collapsing costs, so productivity beats inflation. If it holds, rates can stay low for long, even with an inverted curve.
The most partisan macro voice in the market, with a record of misses. Read her as a framework and a counterparty, not a forecast.
What would change my mind
AI spending growth slows and credit spreads still don't jump, and commitments like $50B per GW turn into realized returns. If the trigger never flashes, her bet was right.
How to read this
The speaker is ARK Invest's founder and CIO, and ARK's whole thesis is a deflationary technology boom, high real growth and buying disruptive innovation. Every dial in this episode is turned to what suits ARK best (growth up fivefold to 15%, oil back to $30, inflation turning negative). Read it as an extreme bullish framework, not a forecast; she has made several high-profile deflationary-boom calls that didn't pan out.
AI is a deflationary boom replaying the Industrial Revolution
Global real growth has run about 3% for 125 years. Five innovation platforms will at least double it to 6%, and could even multiply it fivefold to 15% as the Industrial Revolution did.
00:00 · Back to the Industrial Revolution
0:18Greetings, everyone. It is employment Friday, and it's Labor Day weekend here in the US, so happy Labor Day weekend to everyone. This month's In the Know will be a little different: we're going to talk about very long-term charts, because we're trying to make connections between the Industrial Revolution and this technology revolution. This is a work in process for us, so I'll probably be posing as many questions as I have answers. We'll take this journey together and put the pieces of the puzzle together, particularly around inflation and interest rates, and especially now that we know so much more from Fed Chairman Warsh. His Jackson Hole presentation was quite illuminating, and we'll go through that as well.
01:22 · 3% growth for 125 years
1:22I do think one of the reasons economic indicators are transforming is that we are in a technology revolution, and we're going to see numbers that will be quite surprising to the consensus. The consensus view is that nothing much is changing. Real GDP growth globally has been about 3% for the last 125 years. Brett Winton, our chief futurist, developed this chart with help from academic journals. For roughly 125 years, 3% growth globally; the developed world has been slower than that, and China has kept us in that 3% range for the last 25 years.
2:49The chart goes back to 100,000 BC, and of course these are very rough estimates. What it's trying to communicate is that technology revolutions tend to increase real GDP growth by quite a lot. Very few people alive today have experienced anything other than 3% global growth. From 1500 to 1900 there was some innovation, the printing press and so forth, so growth doubled from the prior 1,500 years. But the Industrial Revolution gave us a fivefold increase in real GDP growth: from 0.6% on average during the prior 400 years to 3%.
4:06What we believe will happen in the next five years is that this growth rate at least doubles, and we think that's a conservative estimate, especially seeing the profound growth coming out of everything AI. There are five major innovation platforms evolving today: AI, the biggest catalyst, then robotics, energy storage, blockchain technology, and multiomics sequencing and technology in the life sciences. The Industrial Revolution was more about three major platforms: first the explosion in railroads, then the telephone, electricity and the internal combustion engine.
5:11So, a fivefold increase. Is it possible that we get a fivefold increase in global real GDP growth, to 15%? We think it's actually possible. We know Elon Musk is starting to use 10 to 15%, and anything is possible, especially with Elon, who is driving this revolution in very important ways. The IMF, meanwhile, is expecting 3.1% growth, so our view is more than two times the consensus. And rolling into this year it's still 3.1%: despite the very rapid growth in these technologies, the IMF has not changed its point of view.
02
Rates are normalizing; the yield curve may stay inverted
Real growth lifts short rates while falling inflation holds down long ones. During the Industrial Revolution, before the Fed, the curve was inverted more than half the time.
05:50 · Nominal growth and interest rates
6:02Why am I going through this? It relates importantly to inflation and interest rates. Nominal GDP growth is real GDP growth plus inflation. After we went off the gold exchange standard in 1971, you can see how far nominal growth went into double-digit territory; back then both interest rates and inflation ended in double digits. Since then there has been a secular decline for 40 years: nominal GDP growth and interest rates both fell, so most people in our business had never seen anything but falling rates until recently. COVID brought the last big drop in interest rates, and since then we've been seeing a rise.
7:48We have been saying that inflation is going to surprise on the low side of expectations, perhaps going negative, and I'll show charts on why. But operating on the other side is real GDP growth, which we think will accelerate quite significantly. So there is going to be a tug of war inside nominal GDP: real growth pulling up, inflation pulling down, maybe negative.
8:33In trend terms, nominal GDP growth and the 10-year Treasury yield are pretty highly correlated. The 10-year yield has been basing in a range since 2023, so we're in our third year of this range, while nominal GDP growth, on a 10-year moving average, seems to be breaking out, and one would expect the Treasury yield to follow. That moving average has some endpoint sensitivity. If we're right and real growth goes one way and inflation the other, rates could keep basing around this level. If real GDP growth is north of 7%, into Elon Musk territory, and inflation is only slightly negative, we will see 10-year Treasury yields move up. That's just the market working.
10:03We started at 0% interest rates in earnest in 2008–09, and for a lot of the time after that the Fed was helping rates down. I think in hindsight we'll say that caused big problems and contributed to the chaos going into and coming out of COVID. On an even longer view, where we are today, averaging around 4.4% year to date, is pretty much in the zone from before we went off the gold standard. We were on the gold standard and then the gold exchange standard until 1971, and then all hell broke loose. Now we've corrected that, and we're coming back to a more normal range.
11:31I know that when people hear interest rates are going up, they get very scared of negative ramifications for the equity market. But the equity market is hitting all-time highs even as rates move up. Break long-term yields into an inflation component and a real component, and rates have been rising more because of real growth expectations than inflation expectations. Again, that's the market at work, and we're very happy to see equities holding up so well in the face of rising rates, despite all the talk about the deficit and the $40 trillion of US debt against roughly $30 trillion of GDP.
12:46 · Before the Fed, the yield curve was usually inverted
12:53One more very long-term observation. The Fed was created in 1913, and once we got through the Depression, we've lived in a world where the yield curve has been positively sloped for the most part, unless we were going into a recession. Inverted means long-term rates are lower than short-term rates. Before the Depression, more often than not the yield curve was inverted, and rarely did we see a positively sloped curve. We believe one reason is that we were entering the Industrial Revolution, and the Industrial Revolution had a tendency toward deflation.
14:09We were on the gold standard, but new technologies were having a deflationary impact on inflation. So long rates reflected more of that deflationary undertow, and short rates focused more on growth in the real economy. We think we might be going back to something like that. Before the Depression, the curve was inverted during periods when real GDP was negative, but also during periods when it was not.
15:01We've had our first episode of that. This last go-around we had a highly inverted yield curve, and yet we never went into a recession after the COVID recession. Many sectors did: manufacturing did, housing did, small businesses did, and lower-income consumers have felt like they're in a recession. So the curve had some forecasting ability for certain sectors, but the overall economy got through it. It's the first indication that we might be back in something like Industrial Revolution times, when the curve was inverted more than 50%, I think more than 60%, of the time, with an average inversion of roughly 100 basis points and much deeper inversions at other points.
03
The debt is bearable; money growth isn't inflationary
The deficit fell unexpectedly; government debt to corporate equity is near a record low; M2 is up about 5% a year, and velocity is flattening with labor participation.
15:31 · The deficit and the debt
16:34The deficit has tipped down. Our expectation was that it would not, but it has, partly because defense spending is ramping even more aggressively, and partly because the corporate tax reductions are much more significant than we thought: lots of tax refunds for the building and investment that started last year. Neither is a bad reason. National security is important, and the corporate tax cuts have given corporations huge refunds to reinvest, and they are reinvesting. On balance, if a trend develops, we believe it will be toward lower and lower deficits, reaching the minus 3% that Treasury Secretary Bessent has set as his goal for the end of 2028 (she said "2018"), because real GDP growth is going to be much stronger than anyone anticipated.
17:59Debt as a percent of GDP, going back to the late 1940s, is near an all-time record. Many headlines are screaming about record-breaking debt, $40 trillion; as a percent of GDP it's not quite a record, but it has been hanging in there. The COVID stimulus took us there and we've stayed there, which is very upsetting to a lot of people looking at profligate spending, fraud and waste.
18:47The purple line is government debt to corporate equities. It's not apples to apples: the government doesn't have equities, and this isn't total debt to all equity. But it gives you a point of view. Debt as a percent of equities has been coming down, much as it did in the '90s, when we were in a very good market for equities and, I think, hit our first government surplus in many years. Debt to equity is near record lows, except for the late '90s. What this tells us is that the ability to support the debt has improved, because there has been wealth generation.
20:01That brings into focus the scary talk of wealth taxes. We are adamantly opposed to wealth taxes, mostly because we are so pro-innovation and we think they would destroy the animal spirits. Hopefully that is not where our economy is going, because it would give China an advantage over the US, and China is our biggest competitor in innovation these days.
19:33 · Money supply and velocity
20:49Now fiscal and monetary policy on the same chart. M2 is in green, as a four-year annualized growth rate, and growth in federal outlays is in purple. Compared with the '70s there is no comparison; we're at the lower end of this chart's range, so we can take some comfort. We do agree that as the economy grows faster, federal outlay growth should slow, especially transfer payments and other social payments. And money growth over the last four years is not an inflationary rate. COVID was, but we have undone that, and that is all to the good.
22:03As we've mentioned many times, as opposed to the four-year annualized rate of 1.7%, M2 growth on a year-over-year basis is a little over 5%. If nominal GDP growth is going to accelerate from the five-ish percent it has been running, the demand for money will increase, and we would imagine M2 continues to accelerate.
22:40We're also watching the velocity of money carefully, and it is flattening out right now. The only correlation we've been able to find, trying to explain why velocity started moving down in the late '90s and has trended down since, is the labor force participation rate. Interestingly, that ticked up a bit in today's employment report. If it goes down, with baby boomers retiring and immigrants leaving, we believe velocity would continue to flatten out, if not decline, and we have to take that into account in the supply and demand for money. Some economists disagree. Probably the most important economist to me, Art Laffer, thinks velocity is just a residual. I have watched it over the years and linked it to a number of things, the most important of which is the labor force participation rate.
24:11And again, the yield curve is flattening, moving toward negative territory. We were there in '23 and '24 into '25, all worried about the recessionary ramifications, and we didn't get a recession. Hearkening back to the very long-term chart: we could go negative again, because real growth is accelerating and pushing up short-term rates, while inflation is likely to come down, perhaps dramatically, and long rates will be more affected by that. So long rates could drop below short rates.
04
Inflation will fall hard: oil has peaked, technology collapses costs
Headline PCE at 3.7% is a temporary oil effect from the Iran war. Abu Dhabi left OPEC and raised output 78%; oil could return to $30; sequencing and inference costs are collapsing.
25:00 · The inflation number Warsh is watching
25:14Now inflation. I want to focus on headline PCE inflation because of the Jackson Hole speech Chairman Warsh gave. He is focused on this number, which is at 3.7%, high by the standards of the last 30 years, and he wants it down to 2%. That's a little shift in our thinking, if he really means he's looking only at this measure. M2 is a little over 5% year over year, and you can see the 3.7% next to money growth. If this measure keeps going up, the Fed will tighten more. We think it is temporary, that the oil price impact from the Iran war is the reason, and that it is getting set up to turn down. But he has focused us squarely on it, and we just have to face it.
26:36This one takes the same measure but uses the Dallas Fed trimmed mean PCE, which takes out the tails of high and low inflation. I bring it up because I believe that in his testimony to Congress after he was appointed, he focused on this measure, and it's at 2.3%, very close to 2%. So we don't think he's focused only on the pure headline 3.7%. He has his eye on this one, and probably on true inflation, because one of the task forces he has commissioned is looking at inflation measures, both public and private.
27:42Here's a private measure. It too is closer to 2% on the headline, 2.4%, and core is at 1.3%. So core is starting to telegraph: wait a minute, maybe tightening shouldn't be your next move. Put all three on one chart and you can see what an outlier headline PCE is. Stay tuned on that one.
27:44 · Oil has peaked
28:21Here's one reason we think inflation is going to come down dramatically. Despite two wars, Russia–Ukraine and Iran, oil has not been able to crack its 2008 price, which at the very peak was $147. In this last round with the Iran war we didn't even get as high as the post-COVID price. A few things have happened. Abu Dhabi dropped out of OPEC in May, and its production is up 78% since then, to a record of a little over 4 million barrels per day. Venezuela is threatening to drop out of OPEC, I'm sure encouraged by President Trump, and that's another source of oil that will be developed more aggressively, maybe heading more to the Western world than to China.
29:39A price signal in the $80 to $90 range is a huge incentive for a lot more production, and it is happening. In the US we're up to 13.6 million barrels a day, and we're exporting over 6 million of those; in 2015 we were exporting almost nothing. So we think that behind the Strait of Hormuz a lot of oil is waiting to flood the market. Will Saudi Arabia throttle back? We don't think so, because Abu Dhabi and Saudi Arabia are very competitive. Abu Dhabi has turned its focus to investing aggressively in technology-enabled innovation, and that has been its mindset for at least five years. When I visited, it was clear they were focused on AI well before the ChatGPT moment and were trying to figure out how to capitalize on it.
30:55Saudi Arabia is unlikely to let Abu Dhabi push its oil out, because I believe Abu Dhabi has concluded that the oil price has peaked. Transportation is moving onto the grid, and the grid isn't supported by oil; it's supported by natural gas, nuclear, hydro, solar and wind. So we believe demand for oil has peaked, or is in the process of peaking, however you want to say it, and we've believed that for quite some time. Now the floodgates are going to open, because Abu Dhabi and maybe Saudi Arabia will want to get as much of their reserves out of the ground as they can and capitalize on this price before it falls. We would not be surprised to see oil drop back to $30, roughly the average of the last 50 years. The move away from oil really started 50 years ago, when OPEC quadrupled prices almost overnight after we went off the gold exchange standard. That process has been a long time coming, and I think we're now here.
31:56 · Technology is collapsing costs
32:31There's another reason we think inflation will come down dramatically: technology. Look at how rapidly these costs are falling. In 2003 it cost roughly $2.7 billion to sequence one person's genome; now it costs less than $100, and it's probably going to $10. AI inference costs are dropping 99.99% per year. Both are starting to get into the health care system, and AI into every industry. Productivity is going to be driven importantly by AI, and there's nothing like productivity as a force against inflation.
33:25 · Gold and the Fed
33:35Here is the gold price, and I'm coming back to Warsh. Under Chairman Volcker and Chairman Greenspan, both of whom I believe used gold as their guide, inflation dropped considerably; we beat inflation. Then, starting under Greenspan in the late '90s, we had Long-Term Capital, the Russian default right before it, rolling Asian crises, and finally Y2K. The Fed and monetary authorities around the world feared Y2K would shut the global economy down, because early programmers had assumed we'd be in the 1900s forever. So the Fed eased, and eased again. Rates did go up, but they should have gone up a lot more given the growth and speculation of the late '90s. Then came the tech and telecom bubble, and rates came down during the bust. We had many more easing moves than we should have, and when the history books are written, I believe they'll conclude that this set off a loss of the dollar's purchasing power in terms of gold. We saw it more in gold than in inflation.
35:52Toward the end of Bernanke's term, through Yellen's and into the beginning of Powell's, gold stabilized in a range again. COVID set it off, as did the Iran war; some will say the Russian war and our own confiscation of wealth, with people rushing for an insurance policy against confiscation. So recently it's been fear of inflation after COVID but, perhaps more so, fear of confiscation of wealth. I know Warsh does not like a rising gold price, and it's interesting that gold peaked the day President Trump nominated him. We would not be surprised to see gold stabilize around here, maybe go down; the jury's out, and we're in wait-and-see mode. But if the dollar moves up, as we believe it will because returns on invested capital in the US are rising relative to the rest of the world, we believe the gold price will come down.
05
AI creates jobs, and capital spending has barely begun
August payrolls rose 162,000; companies using AI more aggressively hire faster; non-defense capital goods broke out of a 20-plus-year range after ChatGPT.
37:19 · Jobs: AI will create them
37:17I'll go quickly through the rest of the charts. It's employment Friday, and it was a strong report: 162,000, when I think the expectation was in the 50,000 to 55,000 range. Household employment, which captures more businesses: boom, more than 450,000 jobs created. The average workweek was longer, which means the economy was really cranking in August. On a year-over-year basis employment had been slumping, but we believe it is turning around.
38:01For those who think AI is going to destroy jobs, you've got another thing coming. We think it's going to create jobs, and the Ramp survey suggests that companies harnessing AI more aggressively are growing their employment ranks much faster than others. The labor force has been shrinking. Baby boomers leaving is not going to stop for five years or so, and immigrants leaving is another pressure. We wouldn't be surprised if we're talking about labor shortages in the not-too-distant future.
38:54The unemployment rate is still 4.1%, where it stayed last month, but for the 16-to-24-year-old cohort it went from 8.5% to 9.1%. So entry-level jobs, yes, we're seeing some of those disappear. Again, I'd encourage people looking for work to also go out and start your own business: solve a problem that bothers you, use only AI and don't hire anyone. You will increase the probability of getting a job out there considerably, and you might even be successful with a new business, although only 10% of startups succeed.
38:44 · Consumers, housing and capital spending
39:47The University of Michigan survey shows a little lift in consumer mood, maybe because gasoline prices, though still up year over year, stopped going up. Average hourly earnings growth is 3.1%, still very well behaved; if productivity is in the 2% to 3% range, unit labor costs won't feed inflation. Personal savings are still low, though they ticked up. A lot of people at the lower end of the income spectrum are living hand to mouth, while at the other end very high-net-worth families are enjoying a booming stock market and booming venture funds and are willing to dig into their savings. That's a bifurcation. The yield curve went negative, but consumption held a decent growth rate. This economy has withstood a lot.
41:01Housing has not. Existing home sales, new one-family homes, homes for sale: we thought that inventory line would keep falling, and yet it has turned around. I think builders started building on spec again when interest rates ticked down for a moment; that's the only explanation we can figure out. There are a lot of homes for sale out there, and new home prices are still going down. Existing home prices have been accelerating a little, probably because of high-net-worth buyers. Mortgage rates went down but have come back up, so my guess is builders will want to clear some of those inventories.
42:28Manufacturing PMI: is this sector moving from just-in-time to just-in-case inventories? Probably not. The AI revolution is affecting not just technology capital goods but also non-technology capital goods. Total non-defense capital goods excluding aircraft has broken out since COVID, after 20 to 25 years of topping out in the same range and going nowhere. The growth rate is heading back to nearly the pace of the '90s internet revolution, and you can see how long that was sustained. I would put the marker for this one a little after the ChatGPT moment. So we're only in the first few years of what we believe is going to be years and years of capital spending.
43:42Here's the trade balance. I know the president hates it, but this was predictable. The US is growing faster than the rest of the world; imports rose 3.6% in one month, with some distortions, and exports fell something like 2.5%. If we're going to grow faster than the rest of the world, this deficit will go up, but it will be offset by a capital surplus, as more investors and companies choose to invest in the US as returns on invested capital rise here. Profitability in the United States is very good, which is another reason capital will be attracted here.
06
AI isn't the railroads: revenue is screaming, credit is calm
200 railroads went bankrupt in the 1800s; AI is different: Anthropic pays $50B per GW. There will be creative destruction, but credit default swaps and spreads are calm.
44:00 · Market signals, gold and bitcoin
44:36Now some market indicators. I'm still puzzled by the metals-to-gold ratio: with gold coming down, I thought we'd see a bigger turn, and we haven't, which tells us something may be going on in China limiting the growth of metals. S&P to oil is hovering near its highs; if we're right about oil, it will break out in a very big way from a range that has been in place since the late '90s. S&P to gold: many bears focused on deficits and government debt have been expecting a waterfall here, the S&P falling against gold, looking more like the '70s. We disagree entirely; we think we're moving in the other direction. The gold to T-bill total return ratio went down during the golden age of equities in the '80s and '90s, and we think the same could happen this time, more because of gold than T-bills.
46:36Here's bitcoin to gold. This is promising; it looks like a turn is in. Famous last words, and I don't want these to be famous last words, but it does look like a little bit of a breakthrough. The correlation between bitcoin and gold is very low by historical standards, so seeing bitcoin start to break out against gold is reassuring from our point of view. You know we're big bulls on bitcoin: it's a technology revolution, a new global monetary system, and the first of its kind in a new asset class. We think it has miles to go, and if we're right on inflation, the pressure on gold will be to the downside.
47:26 · AI is not the railroads
47:36We've been thinking carefully about bitcoin and gold and the insurance both provide, because we think this technology revolution is going to put a lot of companies in harm's way, and there will be counterparty risk if enough companies get into trouble. During the 1800s, 200 railroads went bankrupt; there was so much capital chasing the rail opportunity. We don't think that will happen in AI. The railroads were built on a hope and a prayer that a new world would develop and revenues would ultimately come. In AI the revenues are screaming, and the returns on invested capital in this space are huge today.
48:42Elon's terrestrial data centers have turned from huge losses to massive profits, as he pivoted that part of the business, for the short term, into the neocloud business. Huge returns on invested capital. I think Anthropic agreed to pay $50 billion per gigawatt, and I think it cost Elon and team somewhere in the mid-to-high $20 billions. So right there, immediate returns on invested capital.
49:23But we do think there will be a lot of creative destruction as this new world arrives, across robotics, energy storage, AI, blockchain technology and multiomics, touching every one of our sectors. Funds like private credit made huge bets on SaaS. Some companies and stocks are coming back a bit: Salesforce had a very nice rebound after showing how agentic AI and Slack together might benefit it, though it still has a huge legacy software base to deal with. It's not all death and destruction in software. But there will be dislocations in every part of the economy. We've been talking about electric vehicles and robotaxis and their impact on transportation, one big sector that will see a lot of dislocation, but we think every sector will. So where is the counterparty risk in the economy that might accrue to the benefit of both bitcoin and gold? We shall see. We think bitcoin is both a risk-off and a risk-on asset, and in the environment we see, we believe this ratio will keep moving up to all-time highs.
51:14And quickly: no care in the world, according to bank credit default swaps, so no counterparty risk issues yet. High yield versus 10-year Treasuries: again, very low. The credit markets have not been disturbed much by the debacle in SaaS for the private credit funds. There's nothing systemic happening here, and that's a good thing to take away.
51:41 · Closing
51:47This has been a long one, partly because I wanted to set up a letter I'll be putting out in the next few weeks and touch on some of those topics here. Those of you who tune in to In the Know had a preview; you get special treatment. I look forward to the next In the Know, and I wish you once again a lovely long weekend here in the United States, and a lovely weekend in the rest of the world.
Where Indigo landsFurther
Indigo's conclusion
She is the extreme bullish pole and is best read as a named counterparty. The evidence she uses against the bears is the very structure the bears call fragile: Anthropic's $50B per GW is a forward commitment, not a realized return.
What to remember
She shares Konstantine's Industrial Revolution analogy, and both step around something: Konstantine the speed of transition, Cathie the fragility of financing.
Her deflation case rests on two pillars: peak oil and technology collapsing costs, so productivity beats inflation. If it holds, rates can stay low for long, even with an inverted curve.
The most partisan macro voice in the market, with a record of misses. Read her as a framework and a counterparty, not a forecast.
Claims you can check later
Claim
Who
When we will know
How firm
Global real GDP growth doubles to 6% or more, possibly 7% to 15%
Cathie Wood
Next five years
First-hand; extremely partisan
Inflation falls sharply, possibly below zero
Cathie Wood
Future
First-hand; partisan
Oil could fall back to $30; oil demand has peaked
Cathie Wood
Future
First-hand; a call she has made for years
The yield curve may stay inverted for long periods, as before the Depression
Cathie Wood
Future
First-hand; historical analogy
AI creates rather than destroys jobs; labor shortages are coming
Cathie Wood
Future
First-hand; supported by the Ramp survey
No systemic risk in credit markets (credit default swaps and spreads are calm)
Cathie Wood
Now
First-hand observation; a bet against the bears
Back on the long-running theses
adds to
AI capex as a single engine: panic and must-spend are both true “Only the first years of multi-year capital spending, with huge returns” is the most extreme bullish testimony on the must-spend side, hedging the spending panic.
confirms + adds to
Konstantine Buhler, The Cognitive Revolution Two bulls share “the Industrial Revolution replayed”. Konstantine admits the Engels pause and speed; Cathie reads speed as a plus too.
Ray Dalio: Hormuz is a classic test of American power The same strait: Dalio reads geopolitical risk and an oil premium, Cathie a glut of supply waiting to pour in.
What would change my mind
AI spending growth slows and credit spreads still don't jump, and commitments like $50B per GW turn into realized returns. If the trigger never flashes, her bet was right.
Finished. Indigo's take on this piece is in two places:
