Power Lunch - Meta’s Muse Momentum, Qualcomm CEO Interview, The Future of AI 9/23/26

Episode Date: September 23, 2026

The Nasdaq is retreating from Tuesday’s record levels, and the other major averages are falling as the 10-year yield rises to a 19-year high.Brian Sullivan and Kelly Evans are joined on-set by Goldman Sachs’ Brook Dane to recap Meta’s recent run as shares are on pace for their best month in 13 years following the launch of the company’s new AI agent.CNBC’s Jon Fortt joins the sow with a “First On” interview with Qualcomm CEO, Cristiano Amon, to discuss the launch of his company’s latest Snapdragon chipset.Meanwhile, the anchors sit down for an exclusive interview with Former OpenAI employee, Zack Kass, to hear his take on the state of AI safety and his outlook for the industry as both OpenAI and Anthropic gear up for their potential IPOs. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

Government borrowing costs surging to their highest level in nearly 20 years in some stocks. They're feeling the pain. Welcome to Power Lunch, everybody. With Bonne yields popping today, where are the opportunities for you right now? Don't worry, we've got your watch list. As Goldman Sachs' co-head of tech investing, Brooke Dane is here on set. That's not all. The CEO of Qualcomm, Cristiano Amman, also dropping in for a visit.

We'll talk about Snapdragon, chips, and more. And the man who helped take OpenAI's technology from breakthrough to boardroom is here. Zach Cass, their former head of go-to-market, helped build the commercial strategy that brought CHAPT to the enterprise, what he says now about monetizing AI. All right, we have got a big hour ahead. And overall, the overall markets are lower right now. But let's focus on the positive, shall we? And we are positive that lately there's been a lot of investor interest in shares of meta.

The stock is on track for its best month in 13 years. Its new AI push, apparently an investor, Muse. Muse also the name of Meta's just launched new personal AI agent. And in just a few hours, Meta CEO Mark Zuckerberg, will take the stage to tout more on Muse. And he and employees and investors, they have plenty to celebrate. Muse is now the most downloaded app on Apple's App Store. So is meta, making a real case to be the new king of AI? And if so, is now the time to buy the stock.

So that's Brooke Dane. He is head of co-head of public tech investing at Goldman Sachs asset management. Meta, one of his and your team's top picks. Brooke, welcome back. Thanks for having me back. It's great to be here. I mean, how much market cap has meta added in the last couple of days? A couple hundred billion.

Yeah, a huge amount, right? But this is part of the playbook we've seen with AI over the past few years. When you get these breakthrough events, the stocks that are drawn. driving them have persistent, durable alpha, and you get this big re-rating. If you think back over the last 12 months, right, the controversy on meta wasn't, were they investing enough in AI, wasn't was their core business doing well? It was how were they going to open up a new market, you know, from the AI innovation that they were pushing? Mews is the first example of that. And it's the first thing that you can look at and say, ah, I get it. This has product market fit. And look, if there's one thing we know about meta is if they can get user growth, they can monetize that user growth over time. And so, yeah, it's a really positive thing, and it should change the perception and the multiple of the stock.

And so we're very positive on it right now. That said, what if Google matches it? If there's anything we've seen over the past four years, you know, first it's Google and, no, first it's chat GPT, then it's Google, then it's Anthropic, then it's back to Open AI. Now it's about Muse. What happens if the next one comes along

and is able to somewhat match the performance? Yeah, so first of all, they're all going to come out with products like this, right? We've been talking about theogenic roadmap for a long time in both the enterprises, consumer. This is the first one that really actually clicks. But we would... Grockbot, Elon Musk already has what not. If you're on X, Grockbot does exist. Yes. Okay. So this is the first one that's really captured the zeitgeist of the moment.

The question is, will the other products, you know, come and take that share, or will this one be the dominant provider? We don't think we're going to end up in a world where each of us is going to be running 15 different companies' agents. We think you're likely to use one of or two agents or three agents, you know, for specific tasks and have one main one that does most of your work. Is this, listen, they have two billion-ish daily users. So to Kelly's point, does Muse win? Does meta win?

Do they damage, maybe not Google, but maybe OpenAI, is meta the winner? So this is going to be the thing that we're watching and seeing how it evolves over the next two, three years. But right now, this success. is helping to re-rate the stock. The other thing that people should not lose sight of is the usage that you're going to see across these agents is going to drive incredibly high needs for more compute. And if there's one thing we've been saying for the last two years, it's that we're in a compute-constrained environment. We need more chips. We need more memory. We need more optical.

We did some back-of-the-envolve-quick math around Muse. And if you think about very modest success for Muse, you can get crazy amounts of compute that you need to back this up. Can you translate that into stock upside for Intel, AMD, all of which have already been moving, so I don't know how much of that is already priced in? So literally the kind of calculation we were doing, and this is rough math, is looking at, okay, if they had 100 million users on Muse, if they were using 30 or 40 percent of the capacity that meta gives you for free, you know, on 100 million at like 20 or 30, that translated into about 1.6, 1.7 gigawatts of compute needed. Each gigawatt is roughly 50 billion of

CAP-X. Then you've got to do the math of which chip company wins in that chair. But 100 million sort of using this product is not a stretch of a use case, right? Two billion... So you could... So who, what do I buy? Where's the upside? So look, I think, you know, again, entire... The whole chip section that is focused on AI deployments is really interesting, right? So that's the big caps and the ones we really know. Where we think actually right now that there's a really interesting opportunity is actually in the optical layer, which is, you know, we're seeing this product transition shift from, you know, to co-packaged and near packaged optics. We think companies like Lumentum, you know, they can grow

their revenue much faster than the market thinks over the next two years as you have this architectural upgrade, which is requiring more optics across the environment. Are they in the meta ecosystem? And I ask that because do we just want to buy now stocks of companies that if we believe meta is going to be the winner, do we just buy every company that meta is buying stuff from? That is definitely one of the strategies. I would say, though, that to your earlier point, that this is the first one, we don't know what the competitive response is, you wouldn't want to have a portfolio that was strictly just the meta beneficiary, because what if Google's product is amazing? Zucker bust. So, yeah. So, look, we think that you need

diversity across this ecosystem, and we think you've got to pick your spots. The other thing, beyond Muse, just to bring up another topic that I think is super interesting, is what we're seeing with these open weight models. These are the models that companies can run on their own. They can change the weights of the outcomes. They can train them on their own data. That's actually driving a resurgence in the software ecosystem. And one of the things that we've talked about for a long time is the fact that, yes, semis are where a lot of the action, a lot of the inflection and growth is happening. But don't forget about other parts of the ecosystem that can grow.

Question for you before, because that's a, a lot of the way. really, really good point. And by the way, meta was an open source kind of maker as well. So I don't know if they win in that case too. But just going back for a second to Muse, the new question is around who it might disintermediate. So do you, is it too early to have a point of view on all of these internet companies?

Are they, this is very akin to the software discussion. When we first had open claw come on the scene or Anthropics tools, it was software doesn't have a future. This tool is going to push them out. Then it turned into, I don't know how you would describe it, software becomes hyper-enabled by some of these things. Is the same transition going to happen with e-commerce? So in pockets of the economy, you're definitely seeing that already, and the market is telling you we're going in that direction. So we're beginning to look at those winners and losers and think.

Like, my team is using Muse a ton right now. The first most obvious use case. To do what? Lower your bills. So they're sending these agents off to say, lower my monthly subscription in service X, Y, or Z. And it's coming back an hour later and saving them 20 or 30% on their bills. So, like, there is real risk out there that companies that have high recurring subscription charges are going to face pressure because these tools are going to do a good. I mean, you could do it today.

Like, you and I could sit here after the show, pick up the phone and call and, like, you know, negotiate down the fees we're paying to name your service. It's much easier, though, just to tell me this, hey, would you please go? go do this for me, or whatever the next agentic agent is. So to your point, yes, and investors need to be really careful about some of the business models that they may have thought were stable and sticky and high recurring profit margins. Would it be changing? Something like a Goldman Sachs. Have you seen the action in the financials the last couple of days?

Tell me the logic behind these names selling off to the tune of 5, 10%, and obviously investment banking a little bit different, but there's pockets of what you do. So look, I think, first thing, the market has learned with AI risk to shoot first and ask questions later. So what you're seeing in the market these past couple days is a lot of people pivoting on things. In the financial services vertical, I think what people are most focused on is can people optimize their savings rates? And so if you have money sitting in a checking account, not earning much interest, could you use these agents to help you earn higher interest? So the agents pay you more interest, which costs the interest payers more money? Effectively, NIM margins in financials might, you know, change over time.

Again, that is, like, super early down the roadmaps. And it's unclear how many people actually want to turn over real financial information to these services. Yeah, not yet. I'm not ready. The entire research team is apparently giving them access to their banking statements, their credit card. Or they're, you know, they're Netflix bills, right? How are you finding a Netflix bill? Things like that.

All right. Side note, quick, quick thing. The one, when we talk about the growth of data, the one thing, the one thing, the one I'm sure of. Yes. That the amount of data will continue to grow. Is that a fair statement? Very.

The amount of data is already soared in the last decade. That's benefited to sandisks, the sea gates, the Western Digitals. Anybody stores stuff? Yep. Has physical stuff, too, by the way. Everybody I talked to that says, when we talk about the growth of data, they also keep going back to a company called Snowflake, because they say Snowflake is going to be a winner of basically analyzing or aggregating all of this data.

Yep. That's another name that you like. Is the thesis that simple? The thesis on Snowflake is essentially they are one of the leading companies for doing that data aggregation and getting your data in the formats and in readily accessible storage areas. The other thing, though, is open weight models become more prevalent. You can run the open weight models right on top of your Snowflake instance, so you don't need to be using other services to do.

Do it. Save you money, let you have more control over how AI is operating off your enterprise ecosystem. You saw this business, you know, really accelerate last quarter with net new business hitting its highest growth rates in a couple of years right now. And we think it's very sustainable as we sit here. Fascinating. Brooke, always a pleasure. Thank you so much. Really appreciate it. Yep.

Brooke Dane from Goldman. We have a lot more show on tap for you still. OpenAI Sam Altman and Anthropics Dario Amade are set to brief the UN Security Council this afternoon as the annual General Assessian. Assembly is underway. Can the world agree on AI rules or will the technology outrace us all? Former Open AI executives at Kass weighs in on that ahead. But after the break Qualcomm CEO, Cristiano Amman, on the new chips that could bring more AI agents out of the cloud and onto your phones. Stay with us. Welcome back and take a look at chairs of Qualcomm, which are down slightly after the company unveiled its latest mobile chip systems at the Snapdragon Summit. John Ford is here

on set with us, bringing us CEO, Christiano Amand, in a first on CNBC interview. John? Kelly, thank you. Cristiano, good to see you. Wish I was there. You're in Maui right now for that Snap Dragon Summit. And just to put it into context, Qualcomm is the big name at the top of the smartphone, Android, kind of non-apple food chain. And so this summit is a big moment for capabilities. You're talking a lot about AI on device. or at the edge, as you like to say. How do you fit that into the conversations now about AI safety and efficiency?

Well, I think that's actually a great area to be part of the conversation. Look, the main announcement that we have on this Snapdragon Summit right now is actually we have now clarity what an AI smartphone look like. We've been talking about that for a couple of years. There has been a number of technology

in AI that made it possible, and we started to see this become a reality. You have a lot of phones now that are basically AI smartphone. They have agents, and those agents start to operate the phone for you. The way for you to think about this, John, is the phone has been designed for you to operate, you go to the apps, you do things yourself, but then there's also agents. They're going to go to your apps for you. It's going to go to the cloud, and it's going to do things for you on your behalf. It's very interesting engineering challenge. We now have to design chips to allow the phone to be working even when you're not operating it and with an all-day battery life and a lot of AI at the edge. But for those things to be possible, there are two things going to be very, very important.

Trust. You have to trust the agent and the AI is going to be working on your behalf. And then you have to have low friction. Otherwise, you do it ourselves. A lot of interesting conversation. A lot of devices coming up in the market, we announced a brand new chip for this area, including an AI co-processor, which is kind of the same technology we did for our data center business that we're just starting, moving that to the phones. There are companies now asking for 100 billion model parameters running on your device, and some consumers were wanted the AI to be running on their device for their data and to be in control. Okay, so Christian, I got to ask you about the hot new thing in smartphones. You're already in, I believe, the Samsung Galaxy Z-fold, Z-foldate Ultra,

a bunch of other Android phones out there that fold. Now Apple's got a foldable that's coming out in a few weeks. They priced it high. You and I've been talking for a couple of quarters now about the downward price pressure across smartphones, even hitting the premium end a little bit. What do you think happens now? Does this kind of new? wave in foldables, provide some upward pressure, and how are you preparing? I'm just going to tell my own experience. I am probably on my foldable phone number five. I think foldable and flips have been probably part, I think, of the indoor ecosystem for a while. We're very excited that Apple is going to build one. If anything, it's going to drive more.

That form factor, we have seen, you know, the kind of the conversions between mobile and tablets and PCs, that's another example. of that. The mobile market is still kind of depressed because of the memory prices, but we have seen the premium tier have been more resilient. We talk about it in our last earnings that really CO3 fiscal, it was the bottom. We are seeing even with higher prices right now, I think demand is holding. And look, it's the most important consumer electronic purchase of anyone. So it's a very, It's a very capable device and with AI devices are doing much more. So I think it's the kind of the next step on flagship and becoming probably more mainstream. At the other end of the spectrum and kind of at the other end of the chain, data center,

that's an area that you're moving more into with chips. You mentioned it just a bit ago. And I want to ask about a deal that you guys announced a week or two back with AWS having to do some with chips. I believe you're also going to use AWS some more within Qualcomm. And there's warrants involved. Amazon gets some upside if Qualcomm's stock goes up. Now, there have been questions about whether the incentives throughout the industry are structured correctly. So it's about real demand when the revenue shows up. Tell me about this deal, why it's structured the way it is. and if there are boundaries that you have around the kinds of deals,

maybe involving warrants or incentives, that you're willing for Qualcomm to do? Look, I think people may be thinking too much into this. This is, you know, the warrants, it's a very simple. There's actually all within our 10Q, but basically is there is an understanding now that, you know, Qualcomm is a company under transformation. We're going to all the different markets.

Data Center is another one. It did create a great upside opportunity, and it's an opportunity for Amazon to participate into that upside. And actually, if you just look at the difference between our deal and some of the other deals, is the stock price was kind of the market stock price, as you can see in our filings. I think the exciting part here is that Amazon is believing in what Qualcomm can do in the data center is META was our first customer on our two-generation CPU. Amazon is our first customer on custom silicon and some connectivity that we've been working with them.

I look at those things as probably a vote of confidence that will have relevant technology for this space. Cristiano, it's Kelly here. If you don't mind just a quick bigger picture question, you mentioned meta just now. Do you have any kind of personal anecdotes or usage of muse? What do you think the impact is, even just on phone usage, for instance? You would talk about kind of the ecosystem and all of these new apps. If people love them, they're still apps that you're using on your smartphone.

So that and kind of the whole AI supply chain, what impact do you think this is going to have? Kelly, that's a great question. I think that's the first, I think, point I made. I think this whole Snapdragon Summit Summit has been about we now have clarity what an AI smartphone is going to look like. And Muse is just a great example of that. There's a lot of different agents and orchestrators. I've been using News. been one of the, I think, early adopters when they're doing trial.

And I think that's the future. You're going to have agents. They're going to do things for you. And that's how smartphone users is going to change. Cristiano, I appreciate you joining us here as you continue. Snapdragon Summit. Good to see you. Thank you.

Good to see you, too. Thank you. All right. In a way, not a surprise that he's using Muse. Qualcomm has a chip in those, you know, formerly Rayban Metaglux. So, yeah, they've got kind of an AI. Anybody buying those things, by the Google Glass didn't go real well? I used them. I used them. I got them as a gift.

What do you use them for? I said I used it. Like, it's already dead. But apparently they're going to announce it MetaConnect, a version that doesn't have the camera is that right to try to alleviate those security concerns. It's a little creepy. Hair of the glasses with Muse. Imagine the possibility. Imagine people filming you and just walk around. This whole thing is a little creepy, isn't it?

Like, what are we talking? Everything's a little creepy. Everything's a little creepy these things. I'm surprised how many people were willing to take muse and be like, here's access to all this stuff. You know, the Google Glass came out. I made a joke about riding a bike, at college drunk, and I killed a squirrel. I just made that up. And John Oliver was like, there's no way that guy just made that up.

That actually happened. He made like a graphic of me on a bike naked with a beer. Anyway. Congratulations. By the way, it was the thing I'm most famous for. You know what's not creepy? Your new show. It's a special.

Can you tell us about this? The innovation lab. Saturday 4 p.m. If you've got a kid who's a teenager or a young person recently out of college anywhere in that range, or hey, if you're under 30, you're concerned about, is AI going to take my job? So I did a deep dive on this, talked to a number of CEOs who are working on issues behind this, talked to some young people actually out there who are dealing with this right now. So definitely check it out. A lot of data center feelings for sure. but also from the young people, not as much,

but also quite a bit of hope out there. And we've seen the narrative shifting a little bit on what job availability might be. Walter Isaacson makes an appearance in this special as well. Awesome. Great stuff. If only we had some way to show that. Part of the special? No, just like a clip or something.

Well, you have to stay, too. It's a really deep teeth. Maybe on the glasses. Coming up, John, thank you. One rate hike in the books. Is there more to come? Your market may be saying, yeah, and we'll give you ways to invest around it coming up. Welcome back to Power Lunch to NASDAQ right now. Retreating from Tuesday's record high. Stocks are down across the board, NASDAQ down about 1%. Bond yields, by the way, the 10-year, highest yields is 2007.

But your next guest believes higher rates don't necessarily mean stocks fall. Joining us to explain David Waddell, Chairman and Chief Investor Strategist, Coastal Bridge Advisors. yesterday was like our theme was the 90s, which is great. And I invoked 1994-95. In 1994, the Fed raised rates by 2.5%. In 1995, the S&P 500 returned 37.3%. Its biggest return in 30 years. Higher rates did not kill the market.

Then will they kill the market now? They will not. Thanks for coming. Let's recall. Good to see, y'all. Let's recall that recently we went from zero to five and a half percent, and the market went up seven percent over that time period, and we had that drawdown in 2022. So the last seven hiking cycles, the market's been higher, right? As long as it's going higher for the right reasons, interest rates are, if there's growth, then it's just recalibrating.

So if nominal GDP is 6 percent plus, then rates should be going higher, and they are. but if nominal GDP 6% plus, that means corporate revenues are going to be higher, that means earnings are going to be higher, which they are. So we just have to get through this period of digestion. I will say this, being an active student and using, obviously, the Internet for research purposes. If you went back and looked at environments in an annual basis where the tenure was between 4 and 5% and compared it with years where it's between 5 and 6%, the returns were higher in the years where it was between 5 and 6%. So we just have to get comfortable with the concepts I just laid out and we'll go higher from here. Can we just show guys in the back maybe a board with all the treasury yields on them?

It's calmed down in the past hour, but it's been really, really rough day. You have the two-year treasury yield up like 15 basis points today. We know why? Like what happened? We had PMIs. Yeah, well, we didn't have the official ones. That's a funny thing. They were kind of the flash number.

And it was very strong. The overall number was like the strongest in several years. Maybe it started trade. The price pressures, right. But, Brian, look at this. Look at the 10 year. They're number three on your screen, 512, up about 15 basis points today. The five year was up 18 basis points today.

Yeah. Mortgage rates go to 8%. I mean, seriously. I mean, I'm not making a job. No, we're about to see the highest level on mortgage rates. Again, going back several years. You add, I don't know, we're probably at probably 734. I mean, something in that range.

The housing market has already gone flat from this. And if you ask people why, it's, it's, it's, I guess the strong data there were so inflationary pressures and this idea that the Fed might keep hiking. Well, and did they hike rates or did they hike credibility? Because they hiked credibility a lot during that period. So I think, imagine if we had a truce tomorrow, right? And you're a better energy analyst than I. So I don't know where oil would end up, but that would be stimulus.

Like a true truce? Yeah. I mean, oil falls. In how long? 10 bucks, 20 bucks quickly? Yeah, quickly. Right? So after the midterm, maybe that happens, I don't know.

75 a barrel in 90 days. Think about how fast the economy we'd be going if we got oil sustainably down 20 bucks a barrel. So there's a lot of underlying power in this economy. Plus, it's happening worldwide. There's probably stuff going on with the carry trade in Japan. But what you just said is so interesting and weird, and let's dwell on it for one second. Interest rates have gone higher on the back of higher oil prices that are a tax, a break on the economy. That doesn't really make a lot of sense.

Right. And I was actually pleased that when he had the hawkish press release, so it was a hawkish rate hike that we saw rates come down. But I think what's happened is, again, you got at least signals of how strong the economy is. And I'll have to figure out the technicals more when I workshop this tonight. We got one minute left in the interview, and I would never want to go with the allot of time that we have. I would never, ever do that to the team. Okay. What are we buying right now? So I'm actually kind of into what's going on with the muse thing, and here's why.

We've spent all of this time talking about data centers and hyperscalers and frontiers and chips. We have not talked about the application layer at all. So who's actually going to make money on this? The software companies that we thought were going to be killed, it was the SaaS populace. Now it's maybe SaaSTopia because all of those numbers are rallying today. It's the cybersecurity stuff. But it's also those that produce AI agents because maybe instead of us individually producing all of our AI agents that will kill software. Software is actually going to produce the AI agents, and now they have all the free

cash. So we're keeping our eye on it. Would you say that we are in a saskwatch? That's pretty. I knew that. What about a RenaSass? Oh, it's not as good as saskwash. I agree. Which isn't as good as sastopia. There you go. David Waddell. We're not going to go over the time. Thank you. All right. Good to see it. So sassy today. Let's get to the bond market. That's the best I could do. Let's get over to Rick Santelli. He's with Jerome Schneider out in Chicago. Rick. Well, thank you very much, Kelly, and I'd like to welcome Jerome, the boss at Pimco on the short end. Jerome, welcome.

Great to see you, Rick. How are you doing today? Been exciting day, for sure. All right, we've got to go through it. It's all about the ISM PMIs. I was there at 945 this morning. I read the text, so let's set it up this way. Two words they used the positive side, booming, and they called it surging. What did they do on the other side of the ledger?

They said severe supply shocks and backlogs. Kelly's talking about inflation. Backlogs are the very definition of inflation, good and bad, because in the near term, it creates pricing pressures. But in the long term, you have lots of orders to keep the economy humming. Drill down. Yeah, PMI, if you take it on the face value of it is, it's a volatile indicator, but there's one main message to take away today is that there's a broadening effect in the economy. on two factors. Number one, the economy continues to broaden, both from a manufacturing point of view, as well as the services point of view, but also the inflation discussion isn't going away.

And we see that in terms of pricing power, pricing input costs. And so the main takeaway, and really what the market was reflecting to, was a continual data point, which suggests that the inflation discussion isn't going away any time in the near term. And yet, the economies continue to function on a relatively higher growth perspective. Here at PIMCO, you know, we had penciled in a 2% growth number for 2026, there might be some upside risk to that at this point in time. Absolutely. You know, keep it simple. What if we walk away with after the limit? One category is when yields are going up, price going down. If you're holding the debt and yields are going up, that's a problem with hedging. But the upside is, of course, that you're getting more yield on the

equity side. Describe that dualism. Yeah, I think the big main point here is that we're in an environment now where the downside protection from fixed income is exceedingly attractive. We have to really think about this in two fronts. Number one, the starting point to generate income now is above 5% on across the most of the Treasury curve. If you add in some high quality spread, that puts a starting point for income to be 6 to 7%. It's a very different environment in two folds with regard to where we were in 2022 when the discussion of rates moving higher was relatively detrimental portfolios. Today, the starting conditions are much better, and we find ourselves in a world that is a much more balanced approach, even with inflation lingering, to really avoid the potential

downside that we might see in the equity risk pricing. Really said, the equity market has become a lot less forgiving, whereas the fixed income market is now primed to produce income for the foreseeable future. Awesome. You know, Jerome, thank you. We could go on and on, and I'm sure the public would enjoy it. But thank you for being my last guest on a trading floor. I appreciate it. Kelly and the gang, back to you. Thank you, Rick. Rick said telly, Rick, always the best. Thank you very much. Coming up at AI Insider on the push for global cooperation, Zach Kass, who led the go-to-market operation market at Open AI is up next. You're going to want to hear what he's got to say. Huge talks taking place this week between President Trump and Chinese President Xi Jinping.

as she is set to arrive in the U.S. this evening, high up on the agenda will be AI and how to or not to regulate the technology as the two countries are deadlocked and trying to outpace one another. Our next guest, Sam, has some ideas about AI progress, regulation, and more. Zach Cass is the former head of OpenAI's go-to-market

and now an AI advisor and author company founder. Look, let's not get ahead of ourselves. So welcome, first of all, to the show. It's great to have you here. Great to be here. And you obviously know more than most then when you see all of these claims circulating and you see what these rogue agents have been up to in testing environments. What do regulators, what do all of us need to know about the real potential of this technology?

Well, that we don't have time to cover. But let's... Well, thanks for coming home. Yeah, how much time you got. Let's talk about objectively. Whether or not there is a real threat of doom, at this point, the market has... has decided that the chaos is actually quite unnerving. And the threat alone actually probably substantiates some degree of policy whether or not

the actual doom scenario is real. Moreover, the doom scenario, I argue, distracts us from a bunch of other things that we should be focusing on that do merit policy. Like what? Can you be quick? Because I'm already going, well, wait a minute. I don't know about this regulation argument. But tell me.

So there are a number of ways in which AI can harm us. of them have ethical concerns and some of them have very serious, like, legal concerns. A machine that is not clearly aligned to human behavior can do things that bad actors want the machine to do. A machine that is aligned to human behavior but doesn't understand the unintended consequences. For example, if I said to a machine, clear this, make sure that this, this queue of customer support tickets gets resolved today, and the machine just deletes all the tickets. Well, it's technically resolved the queue, but no one's happier. Every I say to an autonomous vehicle, get me home as safe and fast as possible.

And I'm vomiting by the time we get there. It was safe. It was fast. I'm not comfortable, right? Human intent. So there's these other consequences of misalignment that are way, way closer and sort of more imminent than the ambient dread of doing. Can you teach AI ethics or morality? Yes, but it is a much harder problem than you think.

It has technical complications. I'm not even sure you can. Well, here's the problem. You and I, if we could reach an agreement on what values, we want the machine to actually articulate, probably could. Given enough time, we could probably do it. The real issue is that humans don't actually agree on the set of values that we share. And by the way, a lot of people on Earth very clearly share very different set of values.

So what is alignment to one group of people is not clearly alignment to another. This more than ever merits some degree of international policy alignment and consideration. Because we should start to realize that you can try to align a machine till Kingdom come. never actually arrive at a machine. I mean, the idea of trying to regulate a machine to say what we mean when we say resolve a thing this way and say, I mean, it sounds like an almost endless task. Would it not be better to let the market do what the market does, which is if you engineer a product that goes out and causes these problems, then you should be responsible for that product. We should do that too. And if your company has no value as a result,

that's how the market works. We should, I would argue we should do three things. One, we should actually probably pass policy to guard against. catastrophic outcomes. This is a role of government. This is an important role of government. Don't kill us. We want governments to keep us safe. Do no harm. Do no harm. And critically, we want governments, generally speaking, to keep us out of wars and to make sure that catastrophic events don't come our way. And we should pass policy to protect ourselves from catastrophic downsides. And there are ways to do that. Second, we should, as fast as possible, reinforce all of the bad actor policies that exist. If someone uses AI to tear at the fabric of

society, fraud, other kinds of harm, we should prosecute them much more punitively than we do today, right? We let people get away with a lot of defraud, and we shouldn't let them do that. And there should be international alignment to shut down these fraud camps that exist in other places. And the third, and this is important, we should pass product policy laws that are, to a greater extent, do not offer liability to companies that cause clear harm. And this is CNBC. So I think, you know, we talk about the risk of AI, not just to talk about the risk of AI, but as a business news network, as a stock market show, what we want to understand is, do the risks of AI put the big internet spend, AI-related spend at risk? All the money going into

AI is that at risk because people slow down or shut down AI growth? What's saying? you? This is the Chernobyl question. And as a reminder, Chernobyl had a, had a marginal actual loss of life. I mean, we don't know how to many. Thank God, but it was close. Well, this is the point. You're close from evacuating much of southern Europe. If, if you and I had been alive during Chernobyl to decide whether or not policy should be passed to shut down nuclear power, we would have been very forgiven to say, why would we ever build another nuclear power plant? No, not many people on Earth at the time could point to nuclear power as something they would need. There's a parallel here. Most people don't view AI as something they need. Now,

Mews starts to change that equation. We'll come back to that. But most people look at AI and go, I don't want this at all. And if you're telling me there's a material risk, definitely shut it down. And moreover, the Chernobyl problem has this other effect, which is it is very easy to look at this and extrapolate. Look how devastating this could be. And so this is, this is, you could argue, one of these moments where we have a warning, call it the hugging face attack or some other coordinated attack where we get to decide to what extent do we want a policy and some regulation would I think be really good for the market. What kind though?

I mean, I know we got to go, but that's a really hard. So do you build it into training? You, as I said, pass policy that protects the catastrophic downside. And there's an opportunity right now. I wrote a paper with Phil Potter at the Batten School at UVA that's gaining popularity, which basically says that the market is aligning really nicely. China and U.S. have shared interest in actually making this work. No one wants bad actors to do bad things.

There is a ton of technology now that we actually want to distribute to the benefit of the average person. We've got a lot of lead time to actually make up for it. And also, average person wants to know that someone's at the wheel. And the opportunity here is to say, we're going to pass an international alignment standard that says if a machine shows a degree, an ability to do this harm, you cannot release it.

It's a great way to bring it back to the meetings this week. And for us to think about, as they're meeting literally this afternoon with these AI CEOs, if something like that would be on the table. Maybe your paper will be part of that. I think it'll be on the table. There is a shared alignment now. All right.

Zach, thanks so much. Thanks, pleasure. Zach Cass. All right, let's get over to McKenzie Seagallos with a CNBC news update. Hey there, Brian. Ukrainian president Vladimir Zelensky addressed the UN General Assembly moments ago and called the war on Russia ridiculous and never-ending.

It's now approaching its fifth year. Zelensky thanked every country that stands with Ukraine, saying sanctions against Moscow and weapons support for Ukraine are crucial for its fight against Russia. The CDC confirmed one measles-related death in the U.S. today, nearly one month after Pennsylvania first reported deaths to the agency. The federal government and the state have been locked in a dispute over the reporting with Pennsylvania's Secretary of State calling on the CDC last week to publicly recognize four deaths confirmed by the state. And President Trump will reportedly continue his trend of attending high-profile sporting events with an appearance for the president. This President's Cup this weekend at Medina in Chicago. According to front office sports, he plans to go to Sunday's final round.

It will be the fifth golf event he's attended this year, but the first at a golf course he does not own. Brian, sending it back to you. Mackenzie Cigalos, thank you very much. All right, your market navigator is going quantum. After IONQ unveiled a big breakthrough for the industry, we'll navigate it next. Welcome back to Power Lunch. Time now for our market navigator segment. Shares of IonQ climbing today after the company set at test.

the quantum computing industry's first real-time quantum error decoder that was considered a major breakthrough. Lee Munson is the president and CIO over at Portfolio Wealth Advisors, and it's going to try to help us make sense of whether or not this move in some of these quantum stocks is something we can follow through on. Lee, what's your take on the IONQ price action and the related industry response? Okay, I love the price action. I mean, we have seen things dip down from this 15% pop at the open. What you have to remember about IonQ, You're essentially investing in a venture capital project in a public wrapper. You know, we're always, we're always, we're always complaining about how private equity

is keeping these stocks private too long. That is not the case with IONQ. And here's what they've done. They've basically come up with a way, get all these little quantum particles, there's supposed to be two places at one time. It sounds really confusing, but it's not. They're trying to herd a bunch of cats in a room. And they've come up with the technology, this huge bottleneck, where they're going to be

where they have these little gates in place to make sure the cats are where they're supposed to be, whether they're supposed to be a live cat, a dead cat, a yes cat, or a no cat. And that's going to give way to the commercialization of quantum algorithms in a way that some competitors, let's talk about some of those competitors there down, like D-Wave? Like, why is that not following through? D-Wave isn't really the same thing. It's trying to figure out how to optimize some of the compute through sort of a, cooling heating technology. But in the end of the day, IONQ has basically said, we have a leap forward,

we have a technology now that we've gotten rid of a huge roadblock that's going to prevent people from making quantum a reality. But I'm going to tell you, this is still venture. I like cash flow. You know me. I'm one of those old value players. You know, a lot of my clients are retired. We don't like to speculate too much. I think if you're not into the speculation game and you don't want to be in venture, you're going to look at the biggest big companies with cash flow, and that's going to be IBM, that's going to be Google. But for a pure play, IonQ, it's going to give you the biggest bang for your speculative buck. All right, Schrodinger's cat aside. That's the play from Lee Munson on quantum computing. Thank you very much, Lee. We'll see you again soon.

We'll send you right over here. And by the way, keep it right here because we got more power lunch coming back right after this. Your big story in the market is the pop and borrowing cost, but maybe they're going up because oil is going back up after a couple days of decline. WTI crude Kelly up 2% right now to 92-29. There were earlier reports, by the way, that Politico reported the president was considering a 90-day ban on diesel exports. President or the White House saying, that is, as they would say, fake news. So we'll see what ultimately happens.

But crude oil is up today. And as of now, no diesel fuel export ban. On that report, diesel futures fell 6%. And in Europe, they gained 6%. A preview of what could happen if that does come to pass. Quickly take a look at McDonald's. There's Brent, by the way, at 4% today. And McDonald's is down 5%.

Brian, worst performing stock in the Dow. Carl spoke with the CEO today. He said he's worried that inflation will be higher for longer. Which means the price of what? Hamburgers is going to be higher for longer? It means their supply costs are going to be higher for longer. Look at the flash PMIs today. Shrinkflation.

Thanks for watching Power Lunch, everybody. Closing bell starts right. out.