저자: JamieDimon · 외국인
요약 보기
글로벌 경제는 관세·재정적자·인플레이션·지정학 리스크로 복잡하며, 자산가격이 높아 버블권에 가깝다고 경계. 향후 1년 경기침체 확률을 20~30%로 봄. 스태그플레이션 리스크를 강조하고 금리 상승 시 신용 악화 우려. 금은 4%대 보유비용이 있으나 극단적 환경에서 포트폴리오 일부 보유는 합리적이라 언급. 블록체인은 실물 기술로 보지만 비트코인 자체에는 언급 회피. AI는 실질 효과(비용 절감·생산성)가 있다고 평가하고 일자리 감소와 규제 필요성 언급.
다듬은 글
Jamie, you were in New York this morning giving great earnings reports actually. Um, and now you're here. It's the life of the CEO to be on the road. We appreciate your time with us. I just wanted to open with an economic snapshot from your vantage point. What's happening in the economy? What's happening in the global economy?
What are the main signals you're looking at? Yeah. So, first of all, welcome everybody. Thrilled to be here. I have a little langis. I apologize for that. Look, the the global economy is like inscrutable. So, I'm going to tell you uh but I want to point out whenever you talk about it, we always say we're worried about this or uncertain about that.
You can open the newspaper of any month of any year of anything. It's always that way. Always. What you have to do is compare it to an average, you know, and I do think an average is more complicated. It's more complicated because of trade and tariffs. Is more complicated of huge global deficits and inflation is not going away.
the geopolitical situation, the remilitarization of the world, these are very fairly serious things. You know, uh uh interest rates and I think the chance of surprise are pretty high. Underneath that, if you look at America, the consumers doing fine. Unemployment is good. Wages not going up so much.
Labor's gotten a little bit weaker. Wholesale credit's been good, but I suspect that's, you know, we spent we borrowed and spent $10 trillion over the last six years. Of course, that's going to drive corporate profits and and corporate credit. So that may weaken uh the whole world is a deficit issue and so I'm I'm kind of in the cautious category and also may caused by the fact that asset prices are high you know in some form of bubble territory but that doesn't mean they can't go up 20% more credit spreads are historical lows so it looks like the you know the markets world is pricing in a I'm going to call it a continued soft landing and you know if you looked at next year and I hate forecasting what's the odds of a recession well in any one year it's 20 to 30% So, you know, I think the market's pricing in 10, I'd say 20 or 30% is probably more accurate.
The chance of not being surprised by something next year and then you look at factors like tariffs. I look at it's just a straw in the camel's back. You know, it's not the one thing that's going to break the camel's back. It's always a confluence of factors that do that. And it's very hard, you know, maybe AI will one day figure it out, but it's very hard for human beings, you know, to to consistently forecast real inflection points in the economy.
What are some of the longer term cycles that you're keeping an eye on that may not be reflected as much in the numbers today, but you're a little bit worried about? That's a very important point. I always do future back. You know, people look at all the data today. Unemployment today. Well, unemployment was good.
Markets were good in 71 and then they crashed. That was true in 80 and then they crashed. That was true in ' 87 and there was a crash. That was true in 1990 and it got down. So, you can't look at current data. You got to anticipate data. So, here's some anti some anticipatory data. Deficits are not going away and that will bite one day.
I don't know when. Inflation I what I see is health care prices going up 10%. Food prices are not going down. Immigration policies are inflationary. The remilitarization of the world is inflationary. The restructured trade is inflationary. The capital needs from AI and and infrastructure is large growing and probably inflationary in the short run.
So I that's why I look at the risk is that rates go up. they don't inflation goes up and the Fed can't raise cuts rates anymore maybe has to reverse course and I just when I look at risk I just put that in the table among all the possibility of outcomes and we are prepared for that more than most in fact you know if you look at all things being equal all things being equal make more money if interest rates go up on the other hand if interest rates go up because of inflation and there stagflation we will do much worse on credit volumes and a whole bunch of stuff I can't predict can I play a game with What?
Can I play a game? Sure. Can we play overvalued, undervalued? Yeah, you could try. Okay. Um, gold. Let me tell the press you can ask whatever you want and I can say whatever I want. We want to play a game. It's time to play a game. So, gold overvalued or undervalued? I I don't know. I mean, I'm not a gold buyer.
It cost 4% to own it. It could easily go to 5,000 or 10,000 in environments like this. This is one of the few times in my life I said it's semi-rational to have some in your portfolio. Asset price I but asset prices are kind of high. So I'm not in the back of my mind that cuts across almost everything at this point.
Bitcoin I I'm I'm my blockchain is real. Stable coins may have a real use. We have the JP Morgan coin. Tokens will be real. Programmable money will be all that'll be real. I'm not going to say about Bitcoin because then that's all I'm going to read about in the headlines. the next, you know, and then I get, you know, death threats and [ __ ] like that.
So, I just, you know, all that stuff like, so it's not not a pet rock, but maybe a pit rock with TBD, but the underlying technology, you think? Yeah. But I've always said that blockchain is a technology that we've all been yapping about for, you know, for 12 years, and it's not fully deployed, you know, because it's hard to deploy blockchain that we all agree on permissions and rules.
And I do, you know, we do use it. We use it in repo. I think it's going to replace certain uh tech certain systems we all use that are clunky or late or not 247. Those are all legitimate complaints and blockchain may be one of the things that fixes that. Other technologies can fix it too. By the way, it's not the only thing that can fix it.
And sometimes it's a solution looking for a problem. We already move money real-time payments. Most of it moves digitally in effect. You know, it runs through AML, KYC risk systems when it happens. And you can't just replace all that with a blockchain and think you're doing what regulators and governments want you to do.
So, it's a little more complicated than just it's going to replace everything. Next, Nvidia. I I'm I can't comment on them, but uh there's some of these names that I I just don't want to give specific stuff. Unbelievable company. AI itself is real. I know we're going to talk about AI a little bit later, so I'll reserve that for that.
Okay. So, then what about open AI? That's the last one. Well, again, are we going to talk about AI in general? You want to do it now? That's next. the the way I first of all I think you should all use it to stop debating open close little models big it's going to be all that it's going to be faster it's going to get cheaper you have specialized models we we have 2,000 people doing it we've been doing it since 2012 we have hundreds of use cases we can identify benefits of like two billion dollars maybe two and a half I was told today you know either cost saves or revenue enhancements um and it's just part of run how we run the business today just like big math was that before and it's making huge improvements to certain things other things are hard to measure like Gen AI you know we have almost 150,000 people using a week on internal documents for for report writing for reviewing hundreds of you know thousands of legal documents for summarizing things it's just hard to tell the productivity I do think it's real I think the in total AI is real just like if you go back to 1996 in total the internet was real there were a lot of people in there you could look at the whole thing like it was a bubble crash at one point, but out of that came Google, YouTube, Meta.
So, I think in total it'll pay off. It's hard to know exactly which ones will and they won't be the ones you necessarily predict today. You know, there may be some attributes where winner takes all. That may not be true. I mean, we'll we'll see over time how that works. Yeah. I mean, I think that's completely true.
The underlying the hype is a real tech that is extremely transformative. Extremely good. Yeah. And you will get power. It's high capital usage unlike the internet. So it's got different dynamics. The power may not be there in some cases. Bad guys are going to use it, but bad guys use the internet on social media.
They use airplanes for bad things. And so I do think there should be proper thoughtful regulations and guardrails to protect uh the public. It will eliminate jobs. I think people should stop sticking their head in the sand. So did tractors and so did cars and so did and now if it happens too fast we society government and businesses should figure out how how we can save jobs, save people, retrain, reincome assist, you know, early early retirement is something.
So you can't just take all these people and throw them on the street where the next job is, you know, making $30,000 a year when they're making 150. You know that you'll have a revolution. So we should be thoughtful about that, but in the meantime, you should get good at using it and deploy it and we'll figure it out.
So there's so much in there that I could dig into each of it and so I'm going to try and tack a little bit at a time. One is your investment in AI as a company. You know, there's in this well-sighted MIT study that so many of these early um corporate pilot programs of AI have just been giant money pits um and they haven't been that effective in generating real returns for companies.
I know you just came out with a report that said you guys have invested about two billion in you're getting about two billion back. It sounds like break even point finally. Yeah. And I was wrong about the two billion cost. It's considerably less than that. I I think MIT it's been you have spent less I think MIT was talking about generative AI not about AI so you know most of the things that when we apply AI to AML KYC or risk or fraud very specific things or errors or customer service it works reconciliations you apply it you and sometimes by it's just improving your procedures it's not even really AI a it's just part of the process improvement you go in and all of a sudden your headcount's down 40% % and your time to get something done is not 12 days, you know, it's 12 hours or 12 minutes.
So, it absolutely works. You know, you can overount, you know, the the cost saves or the revenues. It works for fraud. We know our fraud costs are coming down. Most people's going up. It works for we spend $10 million a year marketing, you know, and we know it's going to work on marketing. I mean, marketing by its nature has very inefficient, you know, put the stuff on the wall and see what happens, you know, and they'll be able to measure multiple different things.
So uh and it can measure productivity. It it reduces coding time for those who use it by 30%. That's all real. Generative AI is the other category where that's where we're using generative AI to read documents. So you might you know you might have used it to figure out some of these questions here. It's hard to tell like you know what do you say I what never?
Yeah. What are you gonna say? Well it saved me two hours. You know what's that worth? Did you just spend two hours doing something else or so? We don't really know and I I think it's a mistake to try to calculate everything positive. We spend a lot of money getting data into the proper format so it could be used by AI.
We're just doing it. We're not measuring how much it costs. We're just going to get it in, you know, proper database, etc. So, uh, but it is completely real. And I again, I think MIT is either wrong or was just generative AI. And I think it was just generative AI. Interesting. So did you have any big corporate failures though as you were figuring out how to make more efficiencies with AI?
Any learnings way? No, we did something different where we you know we we it's not in tech even though them and tech are like deep partners in everything they do and tech does a lot of the heavy lifting but the AI experts data think of the data CEOs and the data data people are in this area. So and then we have a mirror inside every company you know credit card consumer trading uh and most I think it really works.
Comments 0