# NVIDIA Scandal and AI Market Shifts

**Podcast:** Prof G Markets
**Published:** 2026-11-12

## Transcript

Running a business shouldn't feel like surviving a software group project.
One app for accounting, another for inventory, another for sales, and somehow, none of them talk to each other.
That's where Odo comes in.
An all-in-one business management software that brings every part of your business together.
From sales and accounting to inventory and marketing, all in one powerful platform.
No messy integrations, no bouncing between tabs, and best of all, no spreadsheets.
Stop managing software and start managing your business with one unified system.
Try for free today at odoo.com slash vox.
That's odoo.com slash vox.
Whether you're meeting in person or online, Plot can help you capture the conversations that matter.
Instead of interrupting the conversation to open an app or manually type notes, Plot is designed to be simple and ready when you need it.
With just one tap, you can start capturing a conversation.
No app to open, no complicated setup, and no need to stop what you're doing.
For a limited time, listeners can get 22% off Plot with code PROPGMARKETS at plod.ai slash PROPGMARKETS.
Spelled P-L-A-U-D dot A-I slash PROPGMARKETS.
Money market madness.
If money is evil, then that building is hell.
The show goes on!
The show goes on!
Welcome to Prof G Markets.
I'm Ed Elson.
It is November 12th.
Let's check in on yesterday's market vitals.
The major indices climbed with tech leading the rally.
The AI trade got a boost from Corweave's earnings, which managed to calm investor anxieties for now.
The stock popped 24%.
Meanwhile, the dollar climbed again towards its highest level since April 2025 as traders solidified bets on a rate hike.
And finally, Nvidia shares rose 3% amid news that Jensen Huang has been back at work all week, reportedly trying to project business as usual.
And breaking news, I am recording this from home after production wrapped.
The New York Times has just reported that Jensen Huang had another affair with a direct report within NVIDIA.
And according to the Wall Street Journal, the board is preparing to oust him.
The stock is plummeting in after-hours trading.
Okay, what's happening?
The NVIDIA spy scandal is threatening to complicate U.S.-China relations.
For more than a year, U.S.
authorities have been investigating Megaspeed International, a Singapore-based NVIDIA customer.
Officials suspect Megaspeed of helping Chinese firms get around U.S.
export controls on advanced AI chips.
And Alice Huang, the suspected Chinese spy who had an affair with Jensen Huang, is a former executive of the company.
On Monday, the New York Times reported that Commerce Undersecretary Kessler halted an investigation into Megaspeed after NVIDIA intervened.
And the Financial Times has reported that the Commerce Department is delaying blacklisting Megaspeed despite pressure from several other agencies.
These revelations come as President Trump prepares to meet Xi Jinping in China next week.
The scandal is now raising a much bigger question.
Can America keep its most advanced tech out of China's hands?
Joining us to discuss this, we're speaking with Chris McGuire, Senior Fellow for China and Emerging Technologies at the Council on Foreign Relations.
Chris, incredible week, incredible news that we're seeing here.
There are a lot of places that we could begin, but I think it would be helpful to just get a refresher.
on what our export policy on chips with China actually is and how it's changed.
Because the heart of the issue here isn't really that Jensen Huang had an affair with someone.
The problem is that he had an affair with someone who was allegedly smuggling GPU chips from Singapore into China.
And the question is whether he actually knew about this.
But just remind us, what is our policy right now and how has it changed?
The policy broadly is that advanced chips, really any AI chips, require a license to go to either China or to any China headquartered company anywhere in the world.
And this is the anywhere in the world that has been the case since 2023.
And the purpose is to prevent, say, subsidiaries of Chinese companies that are located outside of China from receiving the chips.
So what this means is for a U.S.
company, if there's a foreign company and you know that they are a Chinese company, then you can't send ships to them.
So obviously, if they're if they are officially a subsidiary of a China headquartered company, then you can't send ships to them.
But even also, if you know that they are a Chinese cutout, you also can't send ships to them.
Now, I think there's valid questions about if this is sufficient, because what this does is it basically makes it lets the private sector do the determination of what is a Chinese company and what is not.
But the concern here is that this company Megaspeed very clearly.
was and is a Chinese company.
It's being led by Chinese spies.
Those spies are trying to buy these AI chips.
They are infiltrating NVIDIA.
And the chips are still flowing to them.
So from a policy perspective, then, the question is, why has this company not been cut off?
Yeah, we'll get to that.
Why isn't the government really investigating into this?
But let's assume that Jensen Huang kind of knew.
what was going on.
That's sort of my position on this.
I don't see how he could have had such an intimate and close relationship with someone, with Alice Huang, and not known that she at least had some connection to the Chinese government.
She was a reporter of a Chinese state media.
He clearly spent a lot of time with her.
Supposedly, NVIDIA executives went over.
to look at a lot of these companies that had been buying these chips, and they found that actually none of the chips were there.
And so they probably could have connected the dots and realized, okay, probably these things are getting sent over to China.
So in terms of government policy, how bad of a breach of policy is that?
Because it's about national security.
And I guess he broke the law if he knew.
I mean, How bad is that?
What is the punishment?
What is the remedy for doing something like that?
Well, with the caveat at first that obviously this is a legal question and people have to make a determination of what people knew, and I'm not making any accusations on that front.
But if you know that a company is a Chinese company, or if you know that a company is engaged in export control violations, then exporting to that company is a felony.
So that and very, very clearly, right, you are violated.
That is as if you are exporting the chips directly to China.
Now, to be clear, I don't know what people knew.
And it's also fair that like knowledge standards are fairly.
This is a kind of squishy thing.
So actually proving that you had the knowledge is something that's fairly hard.
But and this is why maybe this is something that we should have actual more government oversight and less just kind of outsourcing it and then just hoping people comply with the knowledge standard.
But from just from a perspective of what's legal and what's not, if.
If there was actual knowledge that people can prove, that would be a crime.
Like I said, no idea if that's the case or not.
But obviously, like, we are in a situation where those are questions that I think people have to ask on the government side and I think also in the internal independent investigation into NVIDIA's side.
So it's a crime in and of itself.
And then I sort of layer on top of it this national security element, this idea that we are...
I mean, based on our policy in some sort of arms race with China when it comes to AI, and there have been so many concerns about what that might look like and what the implications of that might be.
I mean, are we getting into like treason territory?
Is that going to be a question that is going to be in the minds of legislators and policymakers?
Like, is that what's at stake here?
I would be skeptical that that's kind of what we're talking about.
But the real question here is, is was there sufficient compliance with controls?
Right.
There are clear laws and rules and regulations of what companies can or cannot do.
And if that is violated, then that is a big problem.
I mean, yes, I think there are separate concerns of interactions with a Chinese spy.
What does that mean?
I would I would say is like in all likelihood, I don't doubt that that Jensen didn't.
know that she was a Chinese spy.
You never know.
But like there's even if he knew that like there was a Chinese for the Chinese national and she had ties to China, Megaspeed ties to China, that would be a different situation.
That itself is a problem.
But the way that these espionage cases would typically work is like they don't want you to know that they're a spy, right?
That they're the spy is trying to to extract information from you.
They're probably trying to figure out who has computing power.
They're probably trying to influence Jensen to give more to Megaspeed because.
Jensen's very much the decider on this front, given there's so much more demand than there is supply for chips.
So NVIDIA decides who gets chips.
They want them to go this way.
There probably are some like subtle communications of like, hey, shouldn't the U.S.
and China cooperate?
You know, shouldn't we really work together on AI safety?
I mean, I don't know.
But like that is typically what an espionage campaign would do.
Obviously, then there's valid questions of NVIDIA has been very vocal in terms of their positions on export controls and being pro sending chips to China.
And then the question is why?
And was this something that was influenced by Chinese buys?
I don't know and we won't know, but I think these, you know, now all of this is open for questioning.
And when you talk about what are NVIDIA's positions on export controls in China and why, they just obviously have no credibility at this point.
So the other piece of this is the fact that a lot of government officials knew about this.
And the common center secretary apparently stopped an investigation into this company.
We're now learning that the commerce department is delaying putting Megaspeed on the blacklist, which is surprising because they've been pretty willing and able to blacklist a lot of other companies.
And then Axios also reported that intelligence officials pressed Howard Lutnick to take action against this company since the fall of last year.
And so it's just kind of remarkable that no action was taken until this was...
sort of broken out into the public sphere.
I mean, what are we to make of that?
What does that say about this White House and I guess the extent to which they view this as an actual problem?
There is no real justification for why once the Commerce Department is aware that you have a company that is being run by Chinese spies, that is buying large numbers of AI chips, and the U.S.
policy is AI chips don't go to Chinese companies, that company needs to be cut off.
That's the whole point of the policies.
So I really cannot imagine the rationale for why they should not be cut off.
And it's really stunning that today...
They still can receive chips.
There's no restrictions.
They're still listed on NVIDIA's cloud on these website as a cloud partner of NVIDIA.
It's just unbelievable here.
And obviously you're seeing the administration, other members of the administration, Secretary Besant today, which has been reported multiple times, are pushing the Commerce Department to take action here.
And they're not.
And I think you're right that the Commerce Department in the past has been willing to blacklist entities.
But it's also worth noting that actually recently.
and by recently, I mean for most of this administration, they have not.
It's actually, we are in the longest period between entity listings, between black listings of companies in the last many, many years for commerce, certainly going back decades.
This is a tool that they have just refused to use.
And on top of that, what I'd say is there are actually regulations on the books right now that...
The Biden administration put in place in January 2025.
In May 2025, the Trump administration said they're not going to enforce.
But if they were enforcing those regulations, they would prevent untrusted companies in Malaysia, in Singapore, in a bunch of other countries around the world that are not close allies.
If it's an untrusted company, they can't receive billions of dollars of AI chips without a license, which to me is a very rational policy.
We obviously want companies like Amazon and Microsoft and generally trusted allied companies to be.
building these data centers.
But if you have a company that just pops up out of nowhere and is like, hey, I want five billion dollars worth of chips, maybe the U.S.
government should have a say into whether we should or not.
And we have no visibility into this because the Commerce Department is not enforcing these policies.
So I think very clearly we need to be asking why there's not a good answer.
And at the at the minimum, they should be enforcing the regulations that are already on the books that would actually prevent this situation in the first place.
I think a lot of people.
You asked the question, why?
Why didn't they investigate this?
Or I guess they did, and then why didn't they take any action?
Why did they, in some cases, halt the examination and the investigation, while their own export policy was literally designed to prevent something like this?
There was a policy to stop...
chips going to China.
They learned that the chips are going to China by via some smuggled route through Singapore and Jensen Huang and his relationship with this woman, which they knew about, which the president had been briefed on.
Why would they not do anything about it?
I think a lot of people will think because he's a friend of Jensen Huang.
We know that they have a close relationship.
We know that they speak to each other on the phone a lot.
We know that the president respects Jensen Huang and feels that NVIDIA is an important part of the American economy, which it is.
And so maybe this is sort of a result of cronyism resulting in making exceptions for people.
I mean, is that what we are to make of it?
And if not, why not?
I agree that there is something weird going on.
And what is interesting, though, is the fact that...
You know, you have this reporting that says the Treasury Department is asking the Commerce Department to take action.
The White House is actually pushing the Hill to pass bills that make export controls stronger.
So it seems like, you know, you have the White House is pushing for stronger policies.
You have the Treasury Department, which is generally leading China policy, pushing for stronger policies.
So while I don't.
you know, know what the president has personally said on this, the rest of the indications on the from the administration actually seem to be that they're moving in one direction and commerce is moving in the other.
So for me, I would look into commerce.
What are the ties there?
Why is commerce defending Jensen?
And that, I think, is a question I think investigators need to look into.
I think in particular, like, are there are there are there ties between Cantor Fitzgerald and NVIDIA that that need to be examined, given Secretary Lutnik?
You know, why did Undersecretary Kessler intervene in an like you said, in an ongoing investigation?
into Megaspeed and by proxy NVIDIA and alert NVIDIA about that investigation.
I think something people remember is that the Commerce Department is actually a law enforcement agency.
They have people with guns and badges and who can arrest people for export control violations.
So for a law enforcement official to tell a company that they're being investigated or that their closest business partner is being investigated, like the FBI would never do that.
But that's what's happening here.
So I think there's a host of questions that that not only.
are resulting in extremely bad policies.
But also, like, I think we have to reasonably wonder, you know, is someone benefiting from this?
Is someone profiting from this?
Because I just, I cannot come up with a better explanation for why you would continue to let all these chips flow.
And I should also just say, Megaspeed is undoubtedly the tip of the iceberg.
There's undoubtedly a slew of additional Megaspeeds that are also receiving all these chips.
We'll say more about that because that's going to be so crucial.
for investors especially, because if we learn that there are, I mean, we know that NVIDIA gets a significant amount of its revenue from non-China, Asia, Asian companies.
And if this is another example of, you know, I mean, if this is the tip of the iceberg, then we could be looking at significant amounts of NVIDIA revenue, which is going to be, I guess, voided, shut off.
maybe that they're going to be punished for.
I mean, say more about why you believe that there is more to this story.
Yeah, so there's a few different elements of this.
And ultimately, I do think it's the tip of the iceberg on the smuggling.
I'm actually less concerned about impacts on NVIDIA sales over the longer term.
I'll explain why.
I think that if you look at some of the trade data, there have been $80 billion of AI chips this year that went to Singapore, Thailand, and Malaysia.
They actually were the number two, four, and five importer of AI chips in the world behind the United States.
About half the AI chips that don't go to the United States goes to those three.
countries.
And these are countries that don't have robust AI industries.
Now, obviously, there are many data centers in them, and some of those data centers are run by U.S.
companies.
But there's also a lot of data centers that are run by companies that are not, that are, you know, the cloud providers that are providing all of this compute to China.
So I do think that that has to be looked into because this is very obviously, you know, export control circumvention by China to, you know, gain access to the chips that they can't buy.
But keep in mind that Everyone in the U.S.
and every U.S.
company wants as many of these chips as they can.
And this is the really curious thing for why NVIDIA is so adamant about selling to China and selling to Southeast Asia, because you have also Elon Musk and Andy Jassy out there saying, hey, we can't get enough chips.
We want more chips.
Jensen is repeatedly saying, you know, all our products are sold out.
Everything's sold out in the cloud.
There's way more demand.
So if we're not selling to Megaspeed, those...
Chips are probably going to get bought by SpaceX, by Meta, by Google, by Amazon, by Microsoft.
And now, obviously, you know, there is some limit on demand, but what we're seeing generally is that the demand for chips and data centers and AI compute is voracious.
And the...
Export controls so far, I would argue, have cost NVIDIA exactly zero sales.
Every single chip they have made since their chips started to be controlled has actually still been bought.
They don't have excess supply.
So we still have room to kind of move things away from China or from Chinese cutouts and to U.S.
firms and to allied firms and everybody benefits from that.
Obviously, there's going to be big fallout for NVIDIA generally in their stock about this whole investigation.
But I do think that there's demand slack there from U.S.
companies that.
they could take up if we actually cut some of these flows to the Chinese cutouts off.
Looking ahead to what might happen next, we're seeing on CalShi 20 percent of traders think that Jensen Huang will be out before the end of the year.
What do you think happens next year?
Do you think Jensen Huang stays?
Does he resign?
What's the next chapter?
I'll leave that to NVIDIA to do that internal decision, you know, analysis and investigation and decision.
I mean, you know, they have hired a very serious and professional firm to do the kind of internal oversight, which is good.
I imagine that there is going to be substantial congressional oversight into this, both into the NVIDIA side and into the Department of Commerce and policy side of this.
You know, we'll see where that ends up.
And look, I will say I agree with everybody who says that NVIDIA and Jensen in particular has has done amazing things for this company.
And he has built the company that makes the best products in the world.
They make it better than anyone.
They have incredible chip design capabilities, supply chain operations.
I think it's also very clear that he executed extremely poor judgment in this case and has put the U.S.
national security at risk and is completely up to the investigators to determine what that means.
for the fate of the company.
But, you know, no matter what happens, he obviously has been, you know, an incredible CEO from a company building perspective.
I just think from an efficacy in Washington perspective and kind of the policy approach they've taken, that is now, I think, has very little credibility and they'll have to decide what that means for, you know, shareholders going forward.
Chris McGuire is Senior Fellow for China and Emerging Technologies at the Council on Foreign Relations.
Chris, appreciate your time.
Thanks for having me.
After the break.
AI spending growth starts to slow.
And for even more markets insights, you can subscribe to my weekly newsletter, simply put, at edwardelson.substack.com.
Support for the show comes from Base44.
Starting a business used to mean raising money, hiring a team, finding developers, and spending years just getting the idea off the ground.
With Base44, starting something new has never been simpler.
It's an AI platform where you describe the software you need, and it delivers.
Base44 goes beyond standard prototyping by converting your vision into fully functional software.
It lays out the pages, the data, who uses it, and what the tool should remember.
At its full potential, Base44 becomes your business's operating system, powering everything from day one launch essentials to technology that runs the business.
The best part is Base44 is made to grow with you.
As your business takes off, you can create an effective workflow that supports your needs from selling your products, serving your customers, or running your team.
And you stay in control.
Use it, change it, and keep building.
Go to base44.com slash markets for 30% off the builder plan for your first year.
That's base44.com slash markets for 30% off your first year.
Thanks to Base44 for sponsoring this episode.
Comitea sources coffee from the best roasters in the world, like Intelligentsia, brews it to perfection, and then flash freezes it at negative 321 degrees Fahrenheit with liquid nitrogen to lock in peak flavor and freshness.
It comes on dry ice right to your door in recyclable aluminum capsules.
All you have to do is melt it to make hot coffee, iced coffee, or espresso-style drinks like lattes in seconds.
No machine or plastic brew path required.
The tech behind Carmatea was developed by MIT-trained engineers and leading coffee experts.
Our producer, Alison Weiss, has tried Carmatea, and she says it has made her mornings a lot easier.
She used to spend time making coffee that never quite tasted good.
the way she wanted.
Now she can make a cup in just a few seconds and it actually tastes like something she would get at her favorite coffee shop.
Try it for yourself.
And for the listeners of this show, you can get $20 off at cometeer.com slash profg or use code profg.
That's spelled C-O-M-E-T-E-E-R.com slash profg or use code profg.
We're back with Prof G Markets.
We are days away from one of the most anticipated events in AI, the Anthropic IPO.
Before the trading begins, investors are still trying to get a sense of what the future of Anthropic's business will actually look like.
New data from Ramp's AI Index, which distills transaction data from more than 85,000 American businesses, may do just that.
Ramp economists found that in October, overall AI spend grew less than 2% month over month.
Meanwhile, OpenAI's token spend growth fell to 2.1% and Anthropix fell to 0.5%.
The AI index also shows that overall token spend has been roughly flat since August.
And despite the fact that businesses were using about 30% more tokens in October than they were the month before, they spent only about 10% more to purchase those tokens, which suggests that AI pricing is collapsing.
So what are Anthropix customers telling investors ahead of next week?
Here to break this down, we are speaking with Ara Karazian, lead economist at Ramp.
Ara, thank you for joining us.
Let's dig into your recently published work here on AI from the Ramp AI Index.
I think the thing that really stands out to me is just the overall AI spend month over month, 2% growth, which is just...
significantly lower than what we've seen in previous months when you've been tracking this.
Tell us a little bit about what we saw in terms of token spending and what it says about the AI business at the moment.
Thanks, Ed.
And you're right about the importance of tracking these kinds of spend trends from OpenEye and Anthropic.
The main finding over the last year is that AI spend had been rising.
pretty rapidly between Okanai and Anthropic.
We saw double digit growth month over month for most of 2026.
And that started to slow down over the last couple of months, now down into the single digits for both Okanai and Anthropic.
Now, important to note that it's not a demand issue and that firms are using more AI by pretty much any measure.
They're using more models.
They're spreading their spend across multiple models.
They're using more tokens.
But it's that those tokens have come down so significantly in price because of this price war that their actual spend has remained relatively flat.
Now that's opening a lot of new questions for OpenAI and Anthropic.
The main one being, where do we go from here?
In my post, I outlined some of what I thought were new fronts in the AI trade and ways in which they would have to start competing with each other.
that they didn't necessarily have to a year before.
And that's because there's now new demand from businesses for cost-effective and faster models to support other enterprise use cases outside of coding.
Yeah, just looking at the data that you found on the average price per token.
So for OpenAI and Anthropic, token prices fell almost 15% last month.
The average price per million tokens at Anthropic is 55 cents.
OpenAI, it's 50 cents.
I mean, Is this a story of open-weight models, presumably mostly from China, flooding the market and forcing open AI and Anthropic to reduce prices?
I mean, what is going on?
What is lowering prices as precipitously as we're seeing?
You know, I wouldn't even blame it on the Chinese models.
Not that they don't matter, but in our data set, Chinese models have less than 5% of the market as far as business spend goes.
You know, I honestly think that if you were to ban all the Chinese models tomorrow, you would still see this kind of price competition from OpenAI and Anthropic just on their own.
In addition to the emergence of new American AI labs that are starting to compete with them and also have the advantage of maybe having access to better compute that will then allow them to build better frontier models.
SpaceX AI is a really good example of that.
Both has in its hands a highly performant model, popular with engineering teams, growing in market share, and also has this compute advantage in that they own so many chimps they can rent them out to Anthropic.
That once OpenAid and Anthropic get into some funding crunch and maybe can't invest as much into compute and building new frontier models, who's going to have the advantage there?
It's going to be exactly the model companies that already have access to compute.
Today, that's going to be SpaceX AI.
That's going to be Meta.
They're not big players in terms of AI spend today by businesses.
But you can imagine that they're likely going to go down that road.
They're going to expand their market share there.
And what is today a two-way race between OpenAI and Anthropic is very quickly going to become a four-way race between them and SpaceX AI, Meta, maybe some other players.
That's only going to accelerate the kind of competition and these effects that we've seen so far that have driven down the price of AI.
Just looking at Anthropics specifically, last week you said that their Q4 business spend is on track to increase between 15% and 25% quarter over quarter.
And just for context, the previous quarter, we learned that their revenue grew 47%.
And I believe that before we learned that number, you had put out an estimate for their Q3 revenue growth, which was pretty much...
in line with that.
I believe you had predicted it would be 50%.
By the way, that was a lot lower than investors wanted because they were looking at nearly 150% growth the quarter before that.
So it appears that growth is slowing.
You see it coming in between 15% and 25% next quarter.
Is that still the case based on what you're seeing in terms of spending?
It's a sign of how volatile and dynamic AI spend really is between businesses in that billions of dollars in market share can shift between one company and another within a matter of weeks, not months.
That's in large part because new frontier models introduced by one company will then suck up the demand that is coming from another.
It's also one of the funny things about how the model companies have been reporting ARR in that it is their recurring revenue annualized.
But it's hard to think of any of this spend as recurring when businesses can so easily move it from one provider to another and are already finding ways to cut their spend and control it.
So, look, it's going to be a big challenge for model companies.
We've been tracking it for a while.
I think the IPOs will be at least a good thing for the markets in terms of allowing us to see some more data on the cost structure of these model companies.
But it's one of the great challenges of investing in this kind of market and understanding it as a market participant is how volatile it is.
We haven't seen something like this before.
It seems when I look at your data, when I look at the charts that you guys have published, I have to be clear, like, this isn't the exact numbers.
We're looking, you guys have proprietary access to what businesses are spending.
But so we don't know exactly what the numbers will look like when we see the numbers for Q4.
And then, of course, going into 2027.
It looks as though AI revenue is starting to plateau or AI revenue growth is starting to plateau.
Do you think that is the right characterization of what is happening here?
Do you think that is something that executives at Anthropic and OpenAI are beginning to contend with?
Is that the takeaway?
I think the largest segment of spend in terms of enterprise spend on AI, which is engineering use, is starting to plateau because those are precisely the teams that are incentivized now to figure out how to spend on AI more efficiently.
That's by and large what we hear from our customers on the platform too.
Does that mean AI spend itself is going to flatline forever?
Well, it depends on what the model companies are able to do in terms of their ability to innovate into a new commercially advanced version of AI that drives productivity gains for teams outside of the engineering use case.
Now, there are some potential takers for that growth.
AI for customer support operations, that's the second largest department in terms of spend on our platform that we can observe after engineering.
You can imagine some spend is going to be taken up by that.
You can imagine that accelerating some growth.
Cybersecurity, so AI being used to deploy against cybersecurity attacks, I think will also be another accelerant to that.
But it remains, though.
one to $2 trillion question on the market, which is, how far is this going to go?
Will we have another coding agent revolution?
Because today, many of the projections are hoping for a commercial advancement that has not come true yet.
Yeah, it seems as though in order for this to work now, they need to find new businesses, essentially.
They need to find new areas of business to expand the number because, you know...
Yes, they're generating huge amounts of revenue, tens of billions of dollars, but they're also about to spend hundreds of billions of dollars.
So that top line needs to grow.
This is particularly relevant because Anthropic is about to go public.
We reportedly were going to get a $2 trillion valuation.
That was the target a couple of months ago.
It has since come down to around $1.3 trillion.
We will see what happens.
Do you have any thoughts or predictions on how this will trade and on what investors are going to be looking for when this company finally goes public?
When we talk to investors, there are a couple of things that I think are very interesting and that they want to see from the model companies.
One, for a long time, the expectation was that AI spend would continue to rise in that sort of big spender group, primarily those engineers.
I think the market has started to understand that maybe we can't expect ever rising growth from that segment anymore.
I think the market is hoping to see proliferation and increased intensity of AI adoption, specifically outside the tech sector.
They do want to see where these other bets will go, AI for other non-technical work.
I thought it was really interesting the products coming out for consumers or products coming out.
They're trying to get at sort of office suite.
But again, an open question about whether or not that will be able to power the same rate of growth that we saw, this sort of hyper growth of 2026, is going to have to come from other teams.
It's not just going to come from engineering.
One of the other really interesting things that you have highlighted is the extent to which these two companies really are dependent on spending by a very, very small handful of VC-backed startups.
And the reason I've found this really interesting is because it basically shows us that the AI economy as it stands is basically dependent on money that was raised from investors.
The money that the startups raise from investors in Silicon Valley, that is then sent to Anthropoc and OpenAI.
And essentially, it means that if VC dollars dry up and we're starting to see a little bit of anxiety in the markets, which could signal that the money will start to slow down, then that will spell significant issues for the startups at sort of the bottom of the application layer.
And then, of course, it'll mean problems for Anthropik and OpenAI.
Could you talk a little bit more about what we know about...
how much those VC-backed startups are spending on Anthropic and OpenAI and the extent to which those two labs are dependent on these startups?
It's interesting because people will see that same chart and report it in two very different ways, depending on where they stand on the AI trade.
So the data point that we reported is that the top 1% of companies, 1% of companies rather, just drive 80% of the spend toward OpenAI and Anthropic.
And then within that top 1%, 80% of them, so pretty much all of them, are VC-backed, and they're typically in tech, finance, and technical services.
That is where AI has both proven its value, particularly for those sort of engineering-heavy organizations, and the big spenders are sort of in that VC-backed group.
Now, so many people read that and say, well, that makes sense.
These are the sort of forward thinking companies that are investing so much in AI.
The rest of the market's going to go in that direction.
And as much as I understand that, I do want to remind people that, look, concentration in industry is fine.
It's a problem when it's highly correlated, especially when it's something like VC, where these dollars tend to move with each other, no matter the company or its use case.
And if you did end up in some world in which spend starts to slow down.
Then you see those effects cascade throughout the rest of the stack.
You start to see them disproportionately impact open-end Anthropics revenues.
And if those have a disproportionate impact on the markets, they can affect parts of our economy that don't touch AI at all.
So that really is one of the big risks affecting the AI trade today.
Again, if open-end Anthropics are able to develop a commercially advanced product that drives sufficient ROI.
outside of the tech sector, then, and it drives enough adoption that they're less reliant on these tech companies, then I would start to change my tune.
I wouldn't worry about this at all necessarily.
But that is an if, and it's something that they have not yet necessarily demonstrated.
Aura Karazian is lead economist at Ramp.
Aura, always appreciate your time.
Thank you.
Thanks for having me, Ed.
Okay, that's it for today.
One quick note before we end, our research associate Kristen O'Donoghue is hosting a live stream tomorrow on the 13th at 1 p.m.
Eastern Time.
She will be discussing how America is losing its scientific edge and what that means for our country's future.
To join that, head to profgmedia.com and sign up for Prof G+.
We hope to see you there.
This episode was produced by Claire Miller and Alison Weiss and engineered by Benjamin Spencer.
Our video editor is Brad Williams.
Our research team is Dan Shalon, Christian O'Donoghue and Mia Silverio.
And our social producer is Jake McPherson.
Thank you for listening to Prof G Markets from Prof G Media.
If you liked what you heard, give us a follow.
I'm Ed Elson and tune in tomorrow for a conversation with Catherine Anne Edwards.
The sun shines through the big window and you are the only one buyer.
900€ warm.
That's not a fantasy.
That's life in Magdeburg.