From Tariffs to AI Tech: What’s Shaping the Market Today | Feb. 3, 2025

Kevin Foley: 00:00:02

Welcome to Optimal Insights, your weekly source for real time rate data and expert capital markets commentary brought to you by Optimal Blue. Let's dive in and help you maximize your profitability this week.

Jim Glennon: 00:00:18

Welcome to Optimal Insights, your weekly source for timely market analysis and expert commentary from Optimal Blue. I'm your host, Jim Glennon, Vice president of hedging and Trading client services at Optimal Blue.

Our clients and industry partners have long relied on Optimal Blue for trusted insights and commentary and these podcasts are an evolution of our commitment to keeping the industry informed. Let's dive into today's episode. Welcome everybody. Hope your week's been good.

We're recording this on Friday, but we'll be releasing it Monday when we will be at our conference, our Optimal Blue Summit that we've been talking about on, on this podcast and everywhere for the last few months. I hope to see a lot of you there. We will also do a bonus episode, look out for that kind of midweek from the actual venue.

We're gonna have a studio set up there to do a podcast. So excited about that. But for today we're going to talk with Ben and Kevin.

We'll talk to Ben here in a sec about what's going on in the economy recently. What should we be looking out for next week while we're all at the conference and what are some other current events going on?

And then we'll have Kevin on a little bit later. We'll talk a little bit about of the AI news that's been out there.

I think we all saw the headlines about Deep Seek and there's been some, some subsequent discussions about Nvidia and other, you know, AI platforms and what, you know, what the deep SEQ news means for that and just kind of highlighting some things going on in the broader world of AI. First, maybe a little bit of data. As you know, rates are stubbornly high.

We're about right about six and seven eighths right now in the conventional 30 year if you follow the OBMMI, which again is based on about 40% of locks that are happening in the industry. Volume still stagnant but pretty healthy. We're still at like 80% of pre Covid levels which is, which is pretty healthy.

You know,: 2019

Before we meet up with Ben and talk through some econ info, I just wanted to acknowledge the unspeakable, tragic plane accident that occurred earlier this week in Washington. Just a horrible situation there.

Obviously, our hearts and our prayers go out to those who lost their lives and everyone affected, family and friends of that accident. And it's obviously hard to make a transition from talking about this into our next segment, but we'll, we'll certainly try.

All right, let's welcome Ben. Hey, Ben, good to see you. Appreciate you being here. A lot to talk about economy wise, that we've had some numbers come out this week.

We've also had the FOMC meeting earlier this week. And then we've got, you know, the always important unemployment report next week. Where's, let's start with the, with the Fed meeting.

What are your thoughts on that?

Ben Larcombe: 00:03:06

Yeah, pretty much as expected. No major surprises. Kind of an initial market reaction to some removal in the Fed's statement regarding the phrase progress on inflation.

That phrase was eliminated. And so I think markets were, you know, like, oh, you know, that kind of a little bit concerning.

But it turns out Powell did clarify that it wasn't supposed to mean anything specifically. Really. Just, you know, you can't say progress on inflation forever. Obviously that's, it's kind of a term that is time sensitive.

So I think it makes sense to remove it.

And I don't think there was really anything new in terms of just kind of still running, you know, about a percent above the Fed's inflation target, not necessarily getting any worse at this point. But, you know, progress has stalled and I think everyone's aware of that.

Powell did say they don't need to see the labor market weaken to reach their inflation targets.

So that's definitely a good sign that, you know, we might not need to see the, the weak job market that we thought we might be heading into early last fall.

And it could just be, you know, hopefully inflation comes in line with these slightly elevated rates, kind of dragging that inflation down here in the coming months or maybe the coming year.

Jim Glennon: 00:04:28

Yeah, it was interesting. I did.

It used to be any slight change in language was highly scrutinized and it could potentially move the market, especially in a sensitive spot like we're in. And they did remove that very important line.

And even before Powell made his clarifying remarks in the press conference, the market didn't react heavily. So I think that was good to see. Almost maybe like the market feels like we're at a natural ceiling in terms of rates, mortgage rates.

And then it was also, like you said, good to hear that the Fed acknowledges that they have their foot on the brake pedal right now. So they just need to see inflation probably a bit better for a longer period of time versus having to see something go south like jobs or.

Normally that's when they would start stimulating the economy. But they, they're so hard on the brakes right now that it, it wouldn't be. There's no need to put your foot on the gas pedal.

They're just willing to take their foot off the brake more if inflation starts, continues to remain tempered, even if jobs are still stellar, which would be just kind of a perfect. It's this perfect landing that we've talked about, this soft, perfect landing. Right.

Ben Larcombe: 00:05:35

Yeah, they kind of know it.

I think they expect that, that things will cool on the inflation front over, you know, and it might not happen in the next few months, but it certainly will, you know, in the coming years at some point. And I think they know they've. They're in a good enough spot now, like you said, like, we're at least under 3% inflation. They're on the right track.

And so, yeah, they definitely don't want to risk re. Accelerating inflation by getting too aggressive in cutting. Seem to be in a good spot.

And I think, yeah, heading into it, I, I think that the official tone was like, wait and see. And it seems like Fed officials weren't really willing to lay too much out there in terms of what they thought might happen, etc.

And that's kind of what we got in this meeting is just we're gonna wait and see what happens. We're not in a rush to cut. We're also not really anywhere near a hike. And. And let's just see some more data.

And I think that's what the market expected and that's kind of what we got. And pretty muted reactions in every sense.

Jim Glennon: 00:06:34

Right. And still no. No direct reaction to some of the rhetoric coming out of the White House, which I think is appropriate. Just keeping that separation.

Right. And not taking into consideration even some of the policies that are starting to.

To come out in executive orders, whether it be mess, deportations or a lot of tariff talk, which we'll get to here in a sec because we've had some very recent news on that. So I think that that's healthy as well.

Ben Larcombe: 00:06:58

Right.

Jim Glennon: 00:06:58

All right. So this week we had a couple pretty big numbers that come out. We had GDP earlier in the week and then we had PCE just this morning.

The former was fine. It was a little bit low. Then PCE today pretty much came in on the screws. Right. Any other Any other tidbits out of that? Those releases?

Ben Larcombe: 00:07:16

I think you pretty much nailed it. GDP came in at 2.3% versus 2.6% expected. I don't think it's uncommon to, to miss by a few ten of a percent.

of: 2024

The PC inflation component you mentioned was at 2.5% for the entirety of quarter four, but then the December portion that we just got this morning came out at 2.6% year over year. So a little bit hotter and 0.3% month over month.

The core number there was actually a 2.8% year over year, so still definitely elevated, but again not at anywhere nearly as concerning levels as we were at too long ago. And all those numbers were basically, as you said, right on the screws. No, no major surprises.

Nothing that's going to change any of the current narratives out there regarding inflation.

So kind of another PC report that turned out to be a market non event, which often seems to be the case with, with CPI just getting released a couple weeks prior generally and then other releases such as PPI and even the GDP report. We just talked about some of that, giving some insight into what this morning's PCE release is going to be.

A lot of traders I think, are pretty well able to triangulate what the number is going to look like.

Jim Glennon: 00:08:53

Right. It's kind of old news at that point. Right. Because we've been able to, even though it is the preferred measure for the Fed, it's old news.

It kind of like you said, you triangulate several of those other inflation numbers and then some components of GDP and you can get there or you can see if it's going to be out of line. Good. So speaking of, of inflation and, and the Fed, there's been some breaking news this Friday afternoon. Why don't you lead us on that tariff news?

Ben Larcombe: 00:09:19

Yeah, yeah. So it, it sounds like what we've been hearing for several weeks, if not months at this point.

The Trump White House is going to levy tariffs on Canada and Mexico at 25% and 10% on China.

I think with Canada and Mexico, it's coming from a place of Trump sees some border issues and security and almost as a, a response to what he sees as potential inaction on their part. So it looks like those are actually going to go into effect on Saturday morning.

It's basically certain that we're going to see a retaliation from probably all of those parties. That's generally how these things work. So we'll have to see how the dust settles early next week. I'm sure markets will react further.

We were kind of up on the day on stocks and I know they lost a lot of their gains and I think bonds are selling off a bit as I think a little more inflation is getting priced in. And perhaps the markets are kind of realizing, okay, it's not just talk. This is going to come to be. It's possible we get a quick resolution.

Like we saw this last weekend with the kind of the Columbia, I think was refusing planes of deported immigrants from the US and that created kind of a spat where there were some threatened tariffs and then an agreement was reached and it all kind of came off the board. So we'll see if that happens, if it could just become a kind of a negotiating tactic and what the end result might be.

But certainly kind of entering some uncertainty here. I'll also point out these tariffs are certainly inflationary, but could also be an impact on growth in the longer term.

And we could actually see a bit of a slowdown economically because of it. That could lead to lower rates. No, no guarantees.

But as always with lower rates, you got to be a little bit careful what you wish for and how you're getting those lower rates.

Like I don't think anyone wants to see a job market get leveled and also see lower rates, which we often root for in the mortgage industry because that could have a totally big separate impact as well.

Jim Glennon: 00:11:32

A good effect. Yeah, we're always in that age old conundrum in the mortgage industry where we root for, root for bad jobs numbers.

Ben Larcombe: 00:11:39

Right.

Jim Glennon: 00:11:39

To lower rates. Even though, no, we don't realistically want people to lose their jobs.

It's just, but it is part of the natural cycle and it is hard to tell what part of the cycle we're in now because there's still so many waves that came from the, all the way back to the pandemic and even back to the great financial crisis, if you look at the Fed intervention piece. Right. But anyway, not to. That's a tangent. Yeah. So the, these tariffs go into effect tomorrow.

So by the time you hear this podcast, it's hard to know what the status will be.

Will it be that, that the administration is just looking to get Canada and Mexico to the table, to get them to Agree to some things that are already in discussions or is this something that, that could have a longer term resolution or longer term ramifications or do these tariffs actually go into place and goods that come over the border actually start getting taxed at a, you know, a quarter of their, their wholesale price, which is, is very significant. Yeah. And we'll, we'll have long reaching effects on the, on our economy.

Ben Larcombe: 00:12:39

Right.

Yeah, it'll, I think that next week we'll, we'll probably get some more clarity of whether there's going to be a quick resolution and even setting tariff and other policy. News aside, we do have Fed speakers, job openings on Tuesday and then a non farm payrolls report next Friday.

So it seems like there's a plethora of opportunities for potential market volatility to show up next week and we'll have to see how it all plays out.

Jim Glennon: 00:13:07

Right, so Fed speak on the heels of some of the drama this week and then yeah, the unemployment report, always highly watched, you know. And remember, remember, keep an eye on the components of that.

You know, the last couple of years it's hospitality, health care, construction and government have made up a vast majority of new jobs.

Whereas a lot of sectors, a lot of what you might traditionally call white collar sectors, tech, banking, finance, this sort of thing has been pretty depressed for the past year or so.

So it'll be interesting to see if those, those areas that have been hot continue to stay hot or if we actually start to see those numbers drop into the, into the hundreds or even the double digits of growth.

Ben Larcombe: 00:13:50

Right. I think the, the expectation is going to be right around175,000 again. And, but yeah, dig deeper into those other components.

Yeah, the composition, what are wages doing, what is the unemployment rate doing? And some of those often will tell a little bit different stories.

So you kind of gotta parse through sometimes conflicting evidence to get, get some insights.

Jim Glennon: 00:14:12

That's right. All right, good stuff, Ben. Thank you as always. Appreciate the time. You bet. Okay. Welcome Kevin Foley. How are you sir?

Kevin Foley: 00:14:24

Doing pretty well. Excited to be here. Excited for the conference.

Jim Glennon: 00:14:28

Right. We're all getting ready for the conference. It's late on a Friday.

We're getting last things tied up, but we're really getting into conference mode I think. But still keeping an eye on the news. And lately there's been some kind of earth shaking news as it relates to AI. Right.

I mean AI has been a huge thing in the news for the past year or so anyway. But then this deep seek thing comes out and I still don't fully understand it. I'm not a big reader, so I'm hoping you've read more than I have.

What are your thoughts on, on some of the, the recent news and some of the revelations that have come since then?

Kevin Foley: 00:15:01

Yeah, no, it's, it's a great, great question. Great topic.

And you know, I figured in the same vein that we, you know, try and help originators become more educated consumers of economic news, it might be worthwhile to do a segment like this where we can, you know, help originators become more educated consumers of technology news. Because this is some pretty big technology news that's dropped in the last couple weeks. So, so Deep Seq, what is it? You know, quick recap.

Deepseek is, you know, a company in China and they created a model called deepseek and, and they rolled that out. That was in the last couple weeks. And there are a few things that were very interesting about this.

So number one was that it was on par or better than some of the American based models such as ChatGPT that many of us are familiar with.

And then number two was that it from a processing power perspective, it uses less than 5% of the energy that a model like ChatGPT uses and it can run on significantly less sophisticated chips, which, you know, is important for, you know, the infrastructure of these models so that energy savings translates directly into cost savings.

And so, you know, if you were to go look and use the Deep Seq API, it's significantly cheaper than the American models that are out there, you know, again, by the order of like 95% cheaper, which is huge.

Jim Glennon: 00:16:31

Yeah, it's almost like the math doesn't. Math. Right.

And that's when some of the debate has been the last week or so, because if you go back a couple weeks and you had the announcement from some of the big AI players in the US that they're going to invest half a trillion dollars in AI, then you have Nvidia, you know, the, the, the top player, at least market cap wise in AI worth trillions of dollars. And then Deep Sea comes out and says we did it for 6 million.

Kevin Foley: 00:16:56

And, and so that, that 6 million number, it's, that seems a little bit, that seems a little more debata.

Efficiency gains that the model is seeing do seem to be playing out certainly based on everything that I've read and try and follow some of the folks who are on top of this research and more reliable. But yeah, so it's sort of, I think, to your point, called into question a lot of this Investment from the American, big American companies.

Do we really need this substantial investment in chips and AI technology in order to get to where we need to go, or is it possible to do, to do things more efficiently? So it's been a very interesting sort of dynamic.

The Microsoft CEO, Satya Nadella, he posted recently about the Jevons Paradox, which is this idea that as technology becomes more efficient, it actually increases the demand for that technology.

So instead of potentially having all of this investment dollars that might be wasted on chips and whatnot if they're not really needed, if you can actually be more efficient, the CEO of Microsoft is saying, well, this might actually spur even more investment into the underlying hardware and the technology because the efficiency gains make it a lot cheaper for everybody.

Jim Glennon: 00:18:22

What are some of the controversies though that have come out since some of the news last week and then some of these new revelations? I've read odds and ends, but what do you think is the most, I don't know, maybe has the most legs under it?

Kevin Foley: 00:18:34

Great question.

So there is, if you play around with Deep seq, which, you know, some folks that work for the US government, I know it's been restricted because it's based in China and all these things, but if you play around with it sometimes Deepseek actually thinks that it's OpenAI's ChatGPT. And so it sort of calls into question how much of the training for Deep SEQ was done directly against ChatGPT.

And actually OpenAI came out and said that Deep SEQ has violated its terms of service, by the way, that it may have been training directly against the model.

And so it kind of brings up this interesting question because one could argue as well for OpenAI, for instance, if you were to ask ChatGPT, draw me a picture of a cross between a man and a sponge, it's going to generate a picture that looks very much like a well known cartoon character who is yellow and wears square pants. That's just, you know, the material that it, that it trained on. And so, you know, who owns that intellectual property.

There are all these sort of fascinating questions that are, that are coming out of that. And I think there are, there are sort of two blind spots that led to this coming out.

So two blind spots by the American companies such as, you know, OpenAI and other folks have been very, you know, hyper focused on onshore.

The first is that, you know, ever since ChatGPT was rolled out, there's been a hyper focus on something called AGI or asi, which is artificial general intelligence or artificial super intelligence.

So everyone saw how powerful ChatGPT was and they said, you know, well, wouldn't it be amazing if we just got it to the next level and started finding cures for cancer and solving math problems that humans have never solved before? Well, that focus has sort of taken precedence over trying to make what we have in terms of artificial intelligence more efficient.

And that sort of left an opening, I think, for other players like Deep Seek to come in there and tackle that problem.

The second kind of relates back to the whole point about training against, you know, OpenAI's model, which is that we've, we, the US has, has heavily restricted the flow of advanced chips needed to train these models to China.

But we, we haven't focused on, you know, providing access or we haven't focused as much on, you know, access to these models directly for training purposes.

We sort of left the back door open a little bit by, you know, all you need is a $20 a month subscription, you know, to, to ChatGPT, to, you know, ask it lots of questions.

And you know, not saying that that's necessarily what they did, but it's, it seems like that's, you know, something along those lines may have happened.

Jim Glennon: 00:21:24

Right, so that's the going theory right now I suppose, is that this, maybe you can put together a model much cheaper than what's been built so far if it, if it's taught or it learns from existing models and also you have some access to those, to those very high performance chips that run.

Kevin Foley: 00:21:43

Exactly, exactly. So I got some hot takes, Jim, on what I think might happen. These, these are my own opinions. I, I speak for, for nobody else.

ne, AI will become cheaper in: 2025

We have a model that's again running on, you know, 5% of the, you know, the overall efficiency of some of the major American models and they've open sourced that, which means that they show you exactly how they did it. It's only a matter of time in my mind until some of that stuff starts getting adopted by the major American models. So that's hot take number one.

Jim Glennon: 00:22:28

Number two makes sense.

Kevin Foley: 00:22:30

AI models I think will proliferate. It'll become much easier for any major company who hasn't built their own AI model.

But maybe they have a, you know, the, a big tech presence for them to actually go and create their own models. Because again, now that Deepseek is, is open source, there's sort of a blueprint for how to do that.

So if you don't want to go through another, you know, another model, you could create your own. And there's sort of a roadmap for how to do that. So I know that, you know, like Alibaba and Tencent have kind of recently launched new models.

And, you know, I expect that there's probably going to be more of that sort of thing as opposed to using a model that's already out there. You know, big, big firms, who knows, maybe Amazon might just go in and end up creating their own.

Jim Glennon: 00:23:20

It totally makes sense.

Everything in technology tends to go that way, is that the front runners put most of the money into R and D or any product really, and then you get sort of not so good carbon copies of that, but good enough and much cheaper. The fact that that Deep Seq is much more efficient is a little curious and very interesting because that's. Where did they get that from?

But then from there it really should just proliferate, right where the mom and pop shop can build their own AI model at some point. And then the AI models should make themselves more efficient over time.

I would think you could tell an AI model, make yourself more efficient, go into this code, strip it down, make it faster, make it smarter. Right.

Kevin Foley: 00:24:01

Who knows, Maybe that's, maybe that's what happened. We weren't in the room, I suppose. So it could be something like that.

Jim Glennon: 00:24:07

And they have to have, I mean, Nvidia, Microsoft have to have saw this coming. They've been building technology for many years. They have just seen this coming.

Maybe not the exact way that it was going to manifest itself, but you would think they, they have money and resources geared towards protecting some of that, some of that investment. But again, who knows?

Kevin Foley: 00:24:25

Well, so piggybacking off of that. This, this is my third. This is my final hot take. This is the, this is the last one.

But number three is I believe that the most advanced AI models will become harder and harder. So as I sort of mentioned where OpenAI sort of left the back door open for other models to be able to train against it.

I feel like there's going to be this level of AI that exists today. It's all the stuff that's out there publicly that folks will be able to access and will be able to train against.

But as it gets incrementally better from where it is now, I feel like those are the models, those new versions are going to be the ones that are going to be very tightly locked down.

Because if you open up access to those for everybody or for, you know, large audience, then you kind of risk losing that, that edge and that intellectual property.

So I feel like there's going to be a little bit of a segmentation between the haves and the have nots in terms of the latest, you know, AI innovation.

Ben Larcombe: 00:25:30

Right.

Jim Glennon: 00:25:30

So they're just going to protect their, their IP better than what. What's been done so far and maybe, maybe walk back a little bit of the openness that was supposed to be these AI models.

Kevin Foley: 00:25:41

Yep, yep. Well, we'll see. Again, they're just hot takes. My hot takes, Kevin.

Jim Glennon: 00:25:47

Hot takes. They do not reflect.

Kevin Foley: 00:25:48

They bat about 50, 50. So, you know, it could go one way or the other.

Jim Glennon: 00:25:53

500 is good. Good stuff, Kevin. Appreciate the insight as always.

We'll be watching, you know, more news about this and just generally, you know, we've got our own AI at Optimal Blue. We'll be, you know, learning more about that come next week with some big.

Kevin Foley: 00:26:06

Announcements, a timely discussion, if you will.

Jim Glennon: 00:26:09

Yes. All right, thanks again, Kevin. Have a good weekend. Travel safe.

Kevin Foley: 00:26:14

Awesome. You too, Jim. Thanks.

Jim Glennon: 00:26:16

All right, let's wrap this thing up. Thank you very much, Ben. Thank you, Kevin, for the information, the wisdom, and the conversation. Hope to see you all in San Diego next week.

That's it for today.

Join us next week for another episode of Optimal Insights, where we'll continue to provide you with the latest market analysis and insights to help you stay ahead. Don't forget to follow us on LinkedIn for more updates and to access our latest video episodes.

You can also find each episode on all major podcast platforms. Thank you for tuning in to Optimal Insights.