Steven Feldman on scarcity, debt and choosing real assets well

Key Takeaways

• AI can remain useful while some investments in it disappoint. Feldman separates a data center earning less than expected from one shutting down. ¹
• Demand for power and physical inputs extends beyond AI. Feldman's case also draws on electrification and existing infrastructure needs. ¹
• Scarcity and borrowing are separate parts of the investment story. A durable asset can sit inside a business with difficult financing obligations. ¹
• Owning farmland is different from running a farm. The tenant's costs and ability to pay rent connect the two, and Feldman discloses an interest in a farmland asset manager. ¹
• A lower share price does not establish that a business has recovered. Feldman's discussion of buying a dip turns on the underlying earnings outlook. ¹
• Cash, bonds and diversification appear in his broader discussion of flexibility and time horizons. His comments describe his perspective, rather than a model allocation for every investor. ¹

Key Moments

00:00:13 - AI adoption, pricing and borrowing

00:05:45 - Electricity demand when data centers earn less

00:07:42 - Scarcity and the role of debt

00:11:39 - Farmland, farming costs and rental income

00:17:31 - Berkshire Hathaway and financial flexibility

00:19:26 - Inflation, energy and consumer spending

00:25:18 - Government borrowing and productivity

00:31:28 - Gold, electrification and physical inputs

00:33:07 - Buying a dip and business fundamentals

00:37:34 - Diversification, cash and bonds

A power plant feels a long way from a chatbot. So does a copper mine. For an investor uneasy about the technology boom, businesses built around physical assets can look like a way to depend less on the next AI breakthrough.

Yet the connection can run deeper than the sector label suggests. A power supplier may be counting on new data centers. A landlord may be collecting rent from the companies building them. Both businesses may have borrowed to expand before the hoped-for income arrives.

Steven Feldman, co-founder and CEO of Wealthion, explores that tension in his October 15, 2026, conversation. His interest in scarce assets comes with an important qualification: the land, metal or infrastructure is only part of what an investor owns. The business built around it, and the debt used to finance it, can change the outcome. ¹

Why AI profits and electricity demand can part company

Feldman draws a distinction between a data center earning less than its owners expected and a data center shutting down. A facility that remains busy still needs power, even if competition forces its operator to charge less for computing. As he puts it, “If they're under-earning, they still need electricity.” ¹

The scale of that demand is already visible. In its April 2026 report, the International Energy Agency (IEA) estimated that global data center electricity consumption grew 17% in 2025. Consumption at AI-focused data centers rose faster, by 50%. Those are measures of electricity use, rather than the returns earned by the companies paying for it. ²

That difference leaves room for two things to happen together: customers can keep finding AI useful while some businesses supplying it struggle to earn enough on their investment. Lower computing prices could encourage more use, but they could also squeeze an operator's income. Whether extra volume makes up for a lower price is a separate business question.

The type of spending matters here, too. Electricity supplied to an operating facility is different from equipment ordered for a facility that hasn't been built. A slowdown in new projects could weaken orders for transformers, cooling systems and construction, even while existing data centers keep running. The IEA's April report explicitly linked the pace of future construction to financing conditions and expectations for investment returns. ²

Why essential suppliers can still disappoint investors

There are also customers beyond AI. In its October 6, 2026, outlook, the US Energy Information Administration estimated that US electricity consumption in the third quarter was 4% higher than a year earlier, with a hot summer increasing cooling demand. ³ Copper has similarly broad uses, including electrical wiring, generators, motors, building construction and industrial equipment. ⁴ That breadth gives these industries other sources of demand, along with exposure to weather, construction and the wider economy.

Having more customers doesn't settle how much a supplier earns from them. A regulated utility may need to build lines and substations before it can serve a large new customer, with recovery of those costs governed by approved rates. A power generator selling into wholesale markets faces different pricing risks from one selling under a long-term contract. The amount of electricity sold is only one part of either business.

Who pays for those lines and substations is already before regulators. In June 2026, the Federal Energy Regulatory Commission asked the six regional grid operators under its jurisdiction to justify or reform their rules for connecting large customers and assigning upgrade costs. The initial orders opened proceedings; they did not set new transmission rates. ⁵

Copper producers face their own gap between demand and profit. A higher copper price can lift a mine's revenue, depending on its sales contracts and any price hedges. Running the mine still involves labor, energy, maintenance and investment. If costs rise or production is interrupted, the business may capture less of the higher metal price. And shareholders' returns also depend on the price paid for the shares: a profitable supplier can disappoint if buyers expected even stronger earnings.

How borrowing changes the real asset story

Feldman returns to borrowing because it adds another claim on the income an asset produces. A building can be difficult to replace and useful for decades, while its owner still has a loan coming due next year. Scarcity says something about competing supply. It doesn't say whether the owner has enough cash to meet that deadline. ¹

The sequence is easier to see through a property's rent. The owner pays expenses such as maintenance, property taxes and insurance. What remains is net operating income, measured before interest payments. The mortgage then has to be serviced from the cash available. A larger interest bill can leave less for the owner even if rents and net operating income haven't changed. ⁶

The loan terms determine when that pressure arrives. A rise in market rates doesn't automatically change the interest rate on an existing fixed-rate loan. The effect may come when the loan matures and the borrower seeks a replacement. Floating-rate debt can respond sooner, when its rate resets. ⁷ The Office of the Comptroller of the Currency identifies higher rates, heavy borrowing and near-term maturities as factors that can make refinancing more difficult. ⁸

This is an existing property-market issue, rather than something unique to AI. The Federal Reserve's May 2026 Financial Stability Report, using information through April 23, said inflation-adjusted price indexes based on commercial property transactions had stabilized after significant declines while refinancing vulnerabilities remained. It also described business and household debt vulnerabilities overall as moderate. ⁹

Debt therefore complicates the idea that inflation automatically benefits owners of physical assets. Rebuilding may cost more, but the existing owner's rent may not rise as quickly as insurance, maintenance or financing costs. Owning without debt removes the interest and refinancing obligation; it doesn't remove operating costs or the possibility that the asset's value falls. A lasting need for the asset can coexist with financial pressure on its owner.

What diversification means beyond a sector label

These connections bring the conversation back to portfolios. A technology company, a data-center landlord and an equipment supplier are different businesses. If each depends on the same wave of construction, however, a cut in that spending could reach all three through weaker sales, delayed projects or financial pressure on customers. The connection doesn't mean their returns must move together. It means that counting sectors alone may miss a shared source of risk.

That doesn't make infrastructure a single bet. Some businesses already serve a wide range of customers; others need years of rapid new AI construction to earn the income their owners expect. Diversification can reduce dependence on individual investments, although the Securities and Exchange Commission cautions that it cannot guarantee against losses when markets fall. ¹⁰

A power plant may be far from a chatbot in what it does, yet closely connected in who pays its bills. Moving into real assets can change the business an investor owns while leaving some of the same dependencies in place. On the real-asset question, Feldman distinguishes the asset's scarcity from the business built around it and the debt used to finance it.

What Could Change the Picture

Supporting context from the article's sources.

• AI economics: Computing prices, utilization and customer payments influence the income available to support new infrastructure. ^1 2
• Construction and power connections: Project timing and rules for allocating grid-upgrade costs affect when facilities can begin earning. ^2 5
• Financing: Interest terms, debt maturities and access to refinancing affect how much income remains for an owner. ^6 7 8
• Demand outside AI: Electricity use, industrial activity and construction give power and materials businesses other customers, alongside other sources of risk. ^3 4

Cleaned transcript

Maggie Lake: Steven, we’ve talked a lot about the fact that much of the bull market momentum we’ve seen in equities is built around the AI trade. But we seem to be seeing some signs of trouble brewing there.

Steven Feldman: Yes. I think there have been a few things in the news cycle.

Number one is the price of a token. While adoption has been pretty good, the price you can charge for that adoption is going down.

The second thing is that you’re seeing some cracks in the capital structure. There’s massive borrowing. Of course, Microsoft is going to be able to pay it back, I’m pretty sure. But what made the news in the last couple of days is CoreWeave, a less established company without massive cash flow to cover its capital needs and debt service. The bonds cracked, and they’re trading at something like 70 cents.

The equity story can believe in a lot, but the bond guys don’t have the upside. They just get 100. They’re now saying, “In order to get 100, I’m going to pay 70.”

That is the canary to me. In a very capital-intensive business, there’s a lot of malinvestment, something we’ve discussed before, and now it’s coming home to roost.

Maggie Lake: You wrote a bit about Blue Owl, too. That concerned people when it came to the data centers as well, right?

Steven Feldman: I don’t know if everybody caught that letter or the story in the news, but there was an $18 billion data center being built. Like any large real estate project, it had quite a few approvals and things that needed to be connected.

Oracle was the tenant. They needed a permit to get a pipeline to provide energy, and the permit, shockingly, has been delayed.

Blue Owl put up $3 billion. Lenders put up $15 billion. Oracle sent a force majeure note that basically says, “There’s a reason we won’t be paying rent, but we’re not in default.”

Has that happened before in the history of real estate? Yes. Has it happened on $18 billion projects? Not so much.

Again, I am not in the corner of the room screaming that it’s over, that the trade is over. What I’m saying is that the things we have been discussing on this channel, and what we in the community are concerned about, have now become news items.

Maggie Lake: It’s interesting because I think people will look at some of this and say, “Well, is it isolated?”

What leads you to believe it’s a canary in the coal mine and that there’s something larger that bears watching? Where do you see it bleed over? Why is it connected to the larger investment environment?

Steven Feldman: First, I’ll add a pro, not me. I believe it was Aon, one of the largest insurance brokers. The CEO came out and said the potential losses here are enormous, bigger than people are underwriting.

That’s the entity that is supposed to understand risk the best. There was AIG, which was involved in a financial crisis.

Again, this doesn’t have a straight line. But you have a fragile system. It’s highly leveraged and highly connected.

You see these charts in The Wall Street Journal where Nvidia has become basically the bank, the joint venture partner, the investor and the customer of a group of companies that are leveraging themselves up to the hilt.

They’re probably now as important to the economy as JPMorgan would be. They’ve become an incredibly important player in how this economy is running, and there are signs of trouble.

Does that mean we should all run for the hills and be out of that trade? I don’t think so. But I think it has to be a more discerning, research-based position: Who are the winners? Who are the losers?

If you’ve made money in the trade, I’ve been advocating a rotation out. Take your winnings, and there have been plenty. They’ve come not only from the hot names like Nvidia. If you owned Hewlett-Packard or Dell, you’re up two, three, four times.

It’s okay to sell those. All of the energy has exploded in those stocks. They’re not going to go up four times from here. Take profits and try to find a place to reallocate.

Maggie Lake: I think people increasingly share those concerns because it’s becoming front-page news, with references back to some of the groundwork that looks very similar to the period before the global financial crisis. They want to hedge.

Do commodities need the AI trade? If you’re thinking about going into commodities, because that was the adjacent trade people felt comfortable with, do they need to see that AI buildout? Are they dependent on it happening for that to be an okay place to be?

Steven Feldman: There’s the mathematical part and the emotional part.

The mathematical part says these data centers are being built, and these data centers will depreciate. Whether you’re going to get a high return or a low return, they’re going to continue to get built. You need things to run them.

There’s a big difference between them under-earning and being shut off. If they’re under-earning, they still need electricity. So I think the trade is there.

Now, if there’s a lot of enthusiasm, emotion and momentum in a utility stock, I would say that’s going to come off when the bloom comes off the AI rose. By the way, it’s a “when,” not an “if.”

It will go back to a normal utility-style business. It could be a fabulous business. We could all be using it, but that doesn’t mean every investment will be a good investment.

If those investments turn sour, that will affect the inputs. But on a relative basis, they’re not trading at 50 times earnings. Well, I should say Nvidia isn’t trading at 50 times earnings anymore either.

They’re a more basic component of what is being built. Frankly, in many cases, it remains scarce with or without AI.

We had an electricity transmission deficit in this country before AI showed up. If you’re in the business of providing copper, it wasn’t very long ago that EVs and other things needed the copper. The electrification of everything was requiring copper. Then data centers added to that.

If that were to calm down, yes, there would be less focus on it. But it’s not going away.

Maggie Lake: How are you thinking about what scarcity looks like? If scarcity is the premium, how do you think about moats and profitability for companies? Everyone is going to be able to level up with AI. What does that look like in this new era?

Steven Feldman: There were two questions packed into that.

In an inflationary environment, and we are in one for the foreseeable future, you want to own a scarce asset, but you don’t want to own a leveraged one. I can talk about why I’m not positive how long the inflation lasts, but let’s break it down.

Take real estate as a fundamental case. If you own a building in New York City, or any city, the replacement cost of that building is higher in an inflationary environment. To replace it costs more money. Therefore, your building should be more valuable on a relative basis.

If that building is levered and interest rates are going up, the NOI can go down in the short run, and it could actually be less valuable in the short run.

You want to own something scarce that doesn’t require massive amounts of leverage to be rolling over to keep it going.

That’s sometimes why you’d see the commodity outperform the companies. In the mining space, up until maybe last year, gold completely outperformed the miners because the miners were so capital-intensive that they always had to go into the market.

If rates go up, their debt goes up. Frankly, the equity should go up, too. You need a bigger risk premium on equity. But the gold doesn’t need any of it.

You’re in a market where owning the commodity might be better than owning the operating business. If you do own the operating business exposed to it, you should make sure it’s not levered.

The second piece is an important point on inflation. I’m not making some call that I’m suggesting anybody put a lot of money behind.

I think inflation is a function of a few things. When it went up, it was primarily a function of a war and then energy prices, which flow through the entire economy, especially diesel and gasoline prices.

I don’t know how long that lasts. At some point, it will be over. I’m not predicting when, but at some point it will be over, and that will alleviate some energy pressure.

I think China has such a glut in manufacturing that it will export deflation. It’s going to make things cheaper.

With the exception of memory chips and things that are scarce because they’re in such high demand, I still see quite a bit of surplus in the economy.

If the AI trade dampens, China exports lower-priced manufactured stuff with or without tariffs, and energy comes back down to before the Iran war, you’re going to see inflation actually come down.

You’re going to see the US government interpret it as aggressively down as it possibly can. We don’t experience inflation the same way it’s reported. You pay tuition, you go to a doctor. That stuff seemed to have been going up 10% a year even when inflation was 2% or nonexistent.

That’s the way you get out of the macro debt piece. You’re going to have to have negative real rates. In order to do it, you’re going to have to fudge inflation. But I think inflation is going to come down.

Maggie Lake: I want to circle back to scarcity that’s not levered because I think that answers a question we get a lot: “I hear you on diversifying. I’m interested in increasing my exposure to real assets, hard assets, whatever you want to call them. Something other than tech, because maybe I’m overexposed. But I’m worried about higher interest rates.”

This threads the needle of what you have to look out for. We even spoke to a farmer, and I know you do a lot of farmland, so you know about this. He was making the point that, yes, farmland is valuable, but it depends on what that underlying farm looks like.

Steven Feldman: Real estate, whether it’s farmland or an office building, has three components to it.

Obviously, there’s the revenue line. For farmers, the revenue line is being pressured. Their input costs are higher. They do use diesel.

Fortunately, you’ve seen a really good run in grain prices. Those are up pretty big for the year, actually. Their profit and loss and ability to pay rent on a piece of farmland is a mixed bag.

But if you’re levered and you used to pay 3% and now you’re paying 7%, your levered equity has gone down.

You’re right, I own an asset manager in the farmland space for row crops. It’s called Goldcrest, and it’s unlevered. We did that for a reason.

The whole business is very conservatively managed. It’s not a hot-money trade. It’s a get-rich-slow scheme. It’s a scarcity scheme. You’re not making more land. It’s civilization’s infrastructure: no land, no food, no humans.

It’s had four down years in 50, and down meaning something like 3%. It’s uncorrelated.

If you want to buy it from this year to next year, I’d say probably a hold. If you’re buying it because you’re funding a pension fund, or you want your children to have an asset that’s going to be valuable, then it’s a great asset to own.

This is also a tension we’ve had a lot between how Wealthion thinks about investing and how the average American thinks about investing. Sometimes there’s a bit of a mismatch.

You don’t buy farmland to have the best-performing quarter you’ve ever had in your life. I know I’ve said this on the channel many times: You buy farmland to avoid living under a bridge.

You buy gold to make sure that, if the world doesn’t go the way you think, you have an asset that is a currency proxy, that’s been valuable for 3,000 years and is valuable in every corner of the planet.

I continue to tell people that you’re underexposed to this asset class. That’s part of Wealthion’s mission.

I think one of the real reasons people are underexposed is that it’s not a Wall Street darling. There aren’t enough fees to go around in the business, so there’s not a lot of education.

If people understood it and were better managers of their own portfolios over a long duration, Wealthion wouldn’t need to exist.

We’re here trying to educate and, over time, provide people good ways to get exposure. But if you’re going to throw a dart into a real asset today, just like if you threw a dart into the stock market, you’re going to lose money.

Maggie Lake: This is where nuance matters, and we all have to lean in a lot more.

You wrote about a really interesting sign of the change in these times, and that was Berkshire Hathaway. That was the gold standard of prudent, long-term value investing. You see the change in leadership as indicative of something.

Steven Feldman: The best you can do as an investor is try to pick up on signals and, in the same breath, make sure you’re not biased to your book.

We talk about that in the gold space. We just talked about this in the prep. There are gold bugs out there. They’ll be right when gold goes up, and they’re wrong at many other times. They read every signal as positive for gold. That’s the bug.

I’m not that, and this channel is not that. We look at signals and try to interpret them. With the help of AI, you feed them all in and it gives you a conclusion. No, just kidding.

Berkshire was run by Warren Buffett. If you look at his investing, he was a little late to the party with Apple. Coca-Cola, great franchise businesses, an insurance company, financial services, American Express.

That has had an extraordinary run and may continue because the United States is uniquely situated in the capital markets, in the transfer of payments and with the US dollar being the reserve currency.

But the guy who got the company, Greg Abel, is the infrastructure guy. He was running basically their real assets business. He knows how railroads work, how pipelines work and how power plants work.

You have a company worth a trillion dollars. It’s a bellwether, extremely influential. It’s going to need to prove its worth in a post-Buffett-cult environment, and they picked a guy who’s going to grind it out.

During a period when there’s nothing obvious in the market, he’s going to grind it out in real assets, sitting on $370 billion of cash, waiting for the thing to break.

Just a matter of months ago, I was saying, “Berkshire isn’t that special because, if things break, there are lots of people with money. Apollo comes in. Blackstone comes in.”

But they’re all investing in the AI trade, so they’re going to be licking their wounds, too. They’re not going to be quite as bold when the stock price is down 25%. And there’s going to be Berkshire.

Maggie Lake: It’s incredible because they’ve done it before.

Steven Feldman: Yes. I was saying to myself, “But it’s different.” When they invested in Goldman Sachs during the financial crisis, I don’t think there were a lot of investors who could come in with $5 billion over a three-day weekend.

Maggie Lake: Keeping your powder dry, though, right?

Steven Feldman: I honestly think this will be $50 billion.

Maggie Lake: My gosh.

Steven Feldman: It’ll be saving somebody like Oracle. All the boats will go down, and no one’s going to want the Kuwaitis to own it. No one’s going to want the Saudis to own it.

Maggie Lake: It’s really interesting. I haven’t heard a lot of people say that. You know Warren Buffett thought long and hard about this. It wasn’t accidental that he chose him as his successor. I think it’s a really interesting signal.

Let’s talk a little bit about inflation again because you brought it up before. We just had a week of inflation data hotter than expected, across the board in some cases, double what the expectations were. We know the energy shock is part of it.

You said you expect inflation to moderate. But headline inflation, energy inflation or across the board? You pointed out that not everything is captured.

Steven Feldman: There’s always the inflation question. I wish you could answer it in one sentence.

There are three components. We discussed what you’re measuring. Certain things are pulled out of inflation for official statistics.

Then, of course, what do you experience when you drive into the Sunoco station and try to buy gas? What do you experience when you’re in the grocery market trying to buy a steak?

I went to a steakhouse last night. Steaks in a decent steakhouse in New York have now cracked $100. We talked to Joe, our farmer, about steak.

You say to yourself, “Wow, it feels expensive.” Am I doing the percentage? I don’t know, but something’s amiss.

Even in the beef trade, the agricultural supply chain is filled with diesel, fertilizer and urea. Between the Ukraine war and the Iran war, it’s creating a lot of pressure flowing through the system.

Everything moves in America in a truck. Trucks run on diesel. At my local gas station, I saw $8.85 for a gallon of diesel. I know that isn’t the rate being paid in the middle of the country. I’ve never seen that number.

By the way, I also pay $8.85 for my iced coffee, which I used to pay $1 for. You see it and feel it.

The other part is wealth-effect inflation. You see the margins in US companies and wonder, “I get it if you’re selling memory chips. I understand why your margins have gone up 10 times. But if you’re selling anything else, why are your margins so good?”

The wealth effect has been letting companies continue to charge more and more, and people are still buying.

But certain industries have begun to roll. If you like to be an investor, you have to read the parts of the newspaper you don’t always enjoy. The luxury sector has rolled. LVMH.

There was a moment, not very long ago, when we thought aspirational luxury was unending. Everybody coming from the middle class to the upper class was going to buy a Louis Vuitton bag. But it’s not true.

It turns out it got too expensive. When you have to buy a Chanel skirt for $9,000, people said, “There are only 80 people left who want to do that.” Some other people are losing the thrill. The thrill is gone.

Now you go to decent hotel rooms, and some are $1,000 a night. You say, “My God, it’s unbelievable.”

Maybe if you’re doing great, the stock market is hitting a record every day, you’re a boomer and you’re living the rest of your life, you’ll pay it. But not when the number starts to go down. When your 401(k) starts to go in the other direction, that luxury good rolls.

By the way, I no longer buy iced coffee outside. I gave up. I make it in the house. For $8, I can get something like five jugs.

Maggie Lake: You’ve hit that pain point.

Steven Feldman: I feel I’m being insulted. It costs them nine cents. They put more ice than coffee in the iced coffee, and I’m supposed to spend $9. Then I have to tip. I have to put the milk and sugar in myself, and I have to tip.

At some point, you’re like, “I don’t want to do this anymore.”

That’s also part of inflation. When that cracks, companies have to lower their prices, regardless of what their inputs would require for them to stay in business.

Then there are the things being caused by geopolitical events, which will change. There are two wars. They’re unending. I don’t see the end, but even Afghanistan was over after 18 years. Something will crack.

The pressure on the United States to make these things into peace, with $40 trillion of debt and increasing interest rates because of inflation, has to make something happen. The next thing is much worse.

Maggie Lake: It’s refreshing because I think everyone’s in the inflation camp. It’s easy to see how this never ends. It’s interesting to hear that eventually, whether high prices cure high prices or there’s some change or relief on the energy front, it could ease.

What about the debt issue you talk about all the time? We had Lyn Alden on, and she was pointing out that some of that may fade, but this monetary inflation, this fiscal dominance we’re in, means they’re just going to keep spending at a government level. That’s going to keep inflationary pressure there and be an issue.

Steven Feldman: I talked before about not sounding like a gold bug, but it’s a tale of two cities.

By the way, I think Lyn Alden invented fiscal dominance, and kudos to her, because the world was being run by monetary policy from the great financial crisis: helicopter money, print, print, print, increase the money supply, and that’s how we’re going to solve everything.

That ran out of steam, and it was replaced by $2 trillion deficits. Those deficits are disconnected from politics.

Nobody wants to raise taxes. It feels like political suicide in the United States. Nobody wants to cut spending. That’s political suicide in the United States. “Keep government out of my Social Security,” which is a very famous line.

I don’t know what’s going to solve $2 trillion on $40 trillion. Now you have higher rates, and you have whatever rolls in a year. I understand it’s $9 trillion over a rolling 12 months.

If that’s just up 1%, you’re talking about interest being higher. It’s already higher than defense. It’ll be creeping up on Social Security. There’s this doom loop.

You can say that when the Fed increases rates, it’s only increasing rates at the very short end. That’s great for cash, but the long end is a fiscal phenomenon.

You look at $40 trillion of debt, growing and compounding. You look at some of the risks in the economy. You look at the competition for debt from the AI folks and how much debt issuance there is. People want a little bit more for the risk.

It’s not because of what Kevin Warsh says. It’s because, “Maybe these guys aren’t going to pay it back, or they want to inflate it away, and I need to get a little bit more return for that.”

She’s right. We’re in a fiscal world, and that fiscal world doesn’t want to fix anything.

The other day, I was doing a separate interview, and someone said I’ve been talking about this for a long time and it’s gloomy.

I said, “I am a very open-minded person. If the facts change, my conclusions will change.”

Nobody I’ve run across who comments on my LinkedIn, writes or sends a note into GBI, the metals business, has offered me any solution in the last three years.

All I get is, “It’s very negative.” I say back, “Fine. Just tell me what solves this.”

Are we going to grow out of it because GDP will start growing at 8% because of AI? The IMF just shot that down. Not that they’re the last word on anything.

AI is very productive, but I can only eat three meals a day. You can’t create GDP out of AI. You can just reallocate it.

If everything comes back to what a consumer does, I’m sort of maxed out. I can’t travel any more than I do. I can’t eat any more than I do. I can’t read any more than I do. It’s not like I can consume more books or more TV.

When I look at my Amex bill, I’m maxed out. I don’t know what more I can do to create GDP, even if AI is literally floating me around, making every decision and cooking my meal. I will not eat any more.

I guess maybe it’s just a matter of the US share of global GDP. Maybe because the US is so far ahead in AI, it starts to suck the productivity out of the rest of the world.

In the very short run, that might help the debt. But forget about the world. The world will then be over if you’re going to impoverish 1.4 billion Indians and 1.3 billion Chinese.

There’s a lot of pushback on our ability to do that, so that’s not very plausible.

Maggie Lake: It’s an interesting way to think about it. In that situation, if you follow consumption and the fact that it doesn’t really change the scenario, what does that mean for portfolios?

Does that bring us back to energy consumption? That’s the only thing there seems to be infinite demand for right now.

Steven Feldman: Same-store sales, right? What are the things that exist? What are we going to need more of to run things where the economy is going?

If I could make two trades, I’d make an electricity trade and a monetary trade. There are two things that seem extremely inevitable to me.

One is that we’re in a world of dangerous geopolitical risks, where somebody somewhere can press a computer key and refrigerators in France go to 122 degrees. That story completely freaked me out.

Maggie Lake: It freaked me out, too.

Steven Feldman: Trillions of dollars of debt, already sovereign pressure in Europe. I think it’s inevitable.

For the people saying, “For six months, gold has done nothing,” it’s in a tight trading range, but rates have gone up. I would have thought it would have gone down further.

It’s basically saying, “I’m good. I’ll stick around. I’ll see where the direction goes.”

I think that’s an inevitable trade. The cool thing about the gold trade is that no one should own it for 100%. If you’re wrong, everything else went up.

God bless the thing you use as your insurance policy. It’s like owning a term life policy for 10 years. If you didn’t die, you still won. If you did die, your family got some money.

The next trade is the electrification trade, and maybe cyber protection and protection of everything else. Those, to me, are at the top of the heap.

How do you want to play electricity? Copper, uranium, pipelines. Data centers at some point, I guess, probably not now.

The things that go into the electrification trade are going to remain very important. Even before AI showed up, the EV car was having that conversation. The robots will need to be plugged in. Everything I know now is on a charger.

Then there are battery minerals and the US supply chain of minerals. Very important. We can’t rely on China for these things.

Again, if you have a 100-year duration, which every investor should start with, I would say the agricultural trade is amazing.

Maggie Lake: Scarce, irreplaceable. It goes back to that fundamental point.

Then do your homework or get someone to help you make sure there’s no leverage hidden in those trades. I think that’s a big takeaway.

In terms of mindset, we’ve talked about this, and I think there’s more recognition that diversification is needed. We need to keep challenging what that looks like.

But the buy-the-dip mentality, habit or knee-jerk reaction is so strong. I still see people every day when the Nasdaq goes down asking, not “What should I be in?” but “What’s my entry point on the dip?”

It’s hard to break that.

Steven Feldman: I agree. There’s a fallacy that gets proven to investors time and time again. Buying a bad company or thesis on the dip, you will continue to lose money.

If it’s a true baby-with-the-bathwater situation, yes, I think buying on the dip makes sense.

You hear investors say, “I look forward to one more financial crisis so I can buy.” I wouldn’t be buying Bob’s Chair Company, no disrespect if there really is a Bob’s Chair Company. I would be buying the world’s greatest franchises on sale.

There is some argument for it when it’s a baby-with-the-bathwater situation.

But if it’s an old-school business that feels like a value because it’s trading at three times cash flow, and cash flow doesn’t come back, you’re going to be buying it at 10 times cash flow next year. It’ll look awful.

The idea of “buy the dip” is a silly one. But I get it. It’s in the ether.

We’ve talked about the S&P 500. I’ve thought more about why it remains a very decent vehicle for your exposure.

There was a study that said if you bought individual stocks from something like 100 years ago, 60% of them would either be out of business or you would have lost money. That means 40% of them did something.

But 90% of the return in the stock market was in 77 companies. I could have that number wrong. It could be 79. It could be 83. But it was absolutely less than 100.

It’s very hard to identify them. There have been tens of thousands of companies over the last 100 years, and 77 of them created 90% of the return.

If you want to think about it, and this is the first time I’m saying this out loud, the S&P 500 or the Dow is like a hedge fund without taxes and with low fees.

For three basis points, you could buy the S&P 500 from Vanguard. They’re going to continue to rejigger that thing tax-free to you and try to find those companies.

It’s market-weighted, so it’ll end up with those companies moving forward. It’s not the stock market. It’s a trading system. As I think about it that way, I like it better. Let them do it.

They’re going to let go of the crummy companies. You won’t find Woolworths.

If you took the original Dow 30 and never changed it, you wouldn’t have made any money. If you took the original S&P 500 and never changed it, you wouldn’t have made any money.

What’s happening is they’re actually selling the dip. They’re letting go of companies that went down in price and aren’t doing well. They’ve made a judgment that it’s not coming back, and that’s the end of it.

It’s not the world’s worst thing to decide that somebody else, systematically, is going to help you find the ones that are going to run.

Buying the dip is not the strategy. Finding the winners and diversifying is the strategy.

Every once in a while, there’s a drawdown. If you’re 63 years old, you’re going to retire next year and your IRA goes down by half, what would you pay not to have that feeling?

I say the payment is diversification. That’s the payment. You don’t have to pay anything. You just have to be in things where you can’t have everything go down 50%.

I’m beginning to come back to bonds. I went 10 years telling people the 60/40 mix should be 100/0, or whatever it is, because you got a negative real rate, no real return, and you paid taxes.

You can buy a 10-year bond, get 5-and-something percent, put it in your IRA and you don’t pay any tax. Five percent is not terrible. It’s something.

If the stock market goes down, it’ll stay there, and you can redeploy it. It’s liquid. You can sell it.

Holding cash is fine. Today, holding cash isn’t horrible. You get 4%. If the market were to correct, you could dip into the best names.

I still say 5% to 10% in cash. If you want more in bonds, that’s fine. But try to find companies or assets that are uncorrelated to the S&P 500.

Maggie Lake: Sound advice. Steven, always great to catch up with you. Thank you so much.

Sources

¹ Wealthion, Steven Feldman interview, October 15, 2026.² International Energy Agency, Key Questions on Energy and AI, April 16, 2026. https://www.iea.org/reports/key-questions-on-energy-and-ai/executive-summary ³ US Energy Information Administration, Short-Term Energy Outlook, October 6, 2026. https://www.eia.gov/outlooks/steo/report/elec_coal_renew.php ⁴ Australian Department of Climate Change, Energy, the Environment and Water, Copper and compounds, May 26, 2026. https://www.dcceew.gov.au/environment/protection/npi/substances/copper-and-compounds ⁵ Federal Energy Regulatory Commission, June 2026 grid-connection orders and cost-recovery explainer. https://www.ferc.gov/news-events/news/ferc-launches-aggressive-targeted-action-speed-large-load-integration https://www.ferc.gov/cost-transparency-customer-protection ⁶ Federal Reserve, definition of net operating income, Section 244.14. https://www.federalreserve.gov/frrs/regulations/section-24414-definitions-applicable-to-qualifying-commercial-loans-commercial-real-estate-loans-and-automobile-loans.htm ⁷ Consumer Financial Protection Bureau, fixed-rate and adjustable-rate mortgages, January 14, 2025. https://www.consumerfinance.gov/ask-cfpb/what-is-the-difference-between-a-fixed-rate-and-adjustable-rate-mortgage-arm-loan-en-100/ ⁸ Office of the Comptroller of the Currency, Commercial Lending: Refinance Risk, October 3, 2024. https://www.occ.gov/news-issuances/bulletins/2024/bulletin-2024-29.html ⁹ Federal Reserve, Financial Stability Report, May 2026. https://www.federalreserve.gov/publications/2026-may-financial-stability-report-overview.htm ¹⁰ Securities and Exchange Commission, Investor.gov, Diversify Your Investments. https://www.investor.gov/introduction-investing/investing-basics/save-and-invest/diversify-your-investments

Wealthion is a dba of Greylock Peak Investments, a registered investment adviser. This article is general information only. It is not investment advice or an offer or solicitation, and it is not tailored to your circumstances. Guests’ views are their own and may not reflect those of Wealthion. Wealthion has not verified guest statements, including any performance figures. Investing involves risk, including loss of principal, and past performance does not guarantee future results. Wealthion refers readers to other registered investment advisers and shares in the fees they receive from referred clients, which creates a conflict of interest. Please see our Form ADV Part 2A and consult your own financial advisor before investing.