ROLLUP: Bitcoin on the US Balance Sheet? | Trump's US Crypto Stockpile | Banks Can Now Custody Crypto | Venice AI's Massive Airdrop

Bankless Nation, welcome to the weekly roll-up. Today on the episode, we got Haseeb Kreshi as my co-host for this episode. Haseeb, how are you doing, my man? Good to see you.

Doing well. I'm doing well. How about yourself?

Really good, really, really good. Uh, there's always just a mountain of news to go through, and I'm really happy to have you here, my man. Let's talk about what we're gonna talk about this week. Trump watch. Trump signed a crypto executive order called Strengthening American Leadership in Digital Financial Technology, which included a bunch of things, including a working group for a national crypto stockpile. Hester Peirce and the SEC revoked SAB 121, and which gives, of course, the green light to banks to uh custody crypto. This open source AI platform, Venice, launched this token and made a huge number of people very happy this week, but also some uh drama around the Coinbase listing. Abstract Chain went live with their mainnet, and Gary Gensler gets a new job. Haseeb, before we get into any of this news in specific, how would you give this week a vibe? What was the vibe of this week? Just in a in a sentence.

Chaotic. I think it's been a very chaotic week. Yeah. It's one of these weeks where the outside world is really shaking what's going on in crypto. So it's you're feeling like you're inside the uh the snow globe and um seeing all the stuff getting getting floated up. That that was the vibe this week.

Is that a combination because of NVIDIA and then also politics?

Yes. That's that's kind of I feel like that's gonna be the theme of the year is that a lot of stuff that's not related to crypto is gonna be shaking up the crypto world.

Uh well, both NVIDIA and politics are top of this agenda. So let's go ahead and get right into it. But before we do, a moment to talk about our friends and sponsors. Haseeb, let's get into markets. Let's start with uh NVIDIA. The big headline that I think rocketed around the world is NVIDIA sheds almost $600 billion in market cap, biggest one-day loss in US history. That's 17% off of NVIDIA. This was on the backs of this Chinese AI lab, DeepSeek, that unveiled this new open source model that it claimed only took two months and less than $6 million to build. People are going back and forth on the validity of that claim. But what is basically well understood is that the performance of this new chat GPT competitor is extremely strong and extremely real. Deep Seek got to the first place of the Apple store charts in just a few days, which is pretty crazy. It is currently number one at the time of recording. And then David Sachs, of course, who is the crypto AI czar, says DeepSeek R0 shows that the AI race will be very competitive. I'm confident in the US, but we cannot be complacent. Haseeb, when you saw this news come across your desk, uh, how did you interpret it? What was your reaction?

Honestly, one of my first uh reactions was so I saw the Deep Seek R1 paper when it originally came out uh seven days ago. I guess eight days ago now. So I saw Wall Street had a very delayed reaction to this announcement. So nothing actually really changed between when the paper was was was actually released, even the numbers. So people were talking about the six million dollar model to train. That was actually a predecessor to R1 called um it was it was the their V3 model, and that was released in December.

So it's one of these things where you realize markets are just not efficient.

Like it just took Wall Street people over the weekend to actually understand what the fuck just happened, like what actually hit them. And I think the biggest thing is that there's uh so much misinformation that's out there now about Deep Seek of people who just do not understand

what it means to have,

first of all, how AI labs work and like how much CapEx goes into actually building an AI lab. People really fixating on this idea that it took six million dollars to train.

It took six million dollars to train the final iteration of the model

Right.

in in GPU hours. But how much did it cost to acquire all the GPUs? How much did it cost to pay all the salaries? How much did it cost to do all the other training runs that failed? The reality is that six million dollars.

Uh almost certainly they spent over $100 million on all this stuff end-to-end. Now, to be clear, that's way less than open AI spending, way less than meta spending. That that's very significant because it means that AI is a lot cheaper

than we thought it was going to be. That's good.

To be clear, that's good for you and me. That's good for almost everyone.

AI products should be thrilled.

Exactly. Exactly. It's bad for producers.

Yeah.

Now the Nasdaq is overwhelmingly weighted towards producers because you know NVIDIA is one of the biggest companies in the world now. And yeah, it's really bad for NVIDIA because, in the same way, you know, when oil prices come down,

It's bad for Exxon, but it's good for us. It's good for everybody else. Good for everyone who drives. So I think by and large, this deep sea thing is actually great for all of us. But the reality is that the balance sheets of the Fortune 500, the balance sheets of the NASDAQ, do not contain people spending money on AI

Right.

because

it just hasn't shown up yet as a line item on most people's balance sheets. It's only a cost right now. No one's making money on AI yet. Everyone's paying money to create AIs. But I think in the long run, this is only a good thing.

I mean, this thing is open source. It's literally more open source than Meta is. Like they release all the intermediate models, not just the final model that they released, like they did for Lama. So

DeepSync, I think, is an unalloyed good for the world. Super, super awesome, super bullish for AI generally, super great for the consumer. Um, but just like with anything, you have to overlay it with geopolitics and fear and FOMO and all this stuff. So I I think markets really got this one wrong in having a generally uh bearish sentiment on the entire market and on risk assets. Huge mistake. Definitely right that NVIDIA stock should go down, but absolutely wrong that anything else should make you less bullish on where this technology is going.

Yeah, it would notably a number of crypto AI projects have already integrated the R1, the Deep Seek R1 model. Venice now has Deep Seek R1 in its products. And it's to me, it's a little uh it's I'm happy that we are seeing the creations of these AI labs, which are very expensive, very hard to produce, as you said, tons of CapEx. You know, may like maybe it costs six million dollars to train, but again, like you said, that's built on top of the shoulders of OpenAI and many other people, hundreds of millions of dollars to get to that last six million dollar mile. And then it's open source. And so while open AI is truly closed, and so is Meta and Cloud and all these other you know LLMs, we have this arms race going on between China and A and the United States. But then it if it continually ends up in the hands of the open source community, just you know, three, six, nine months later, and we get all these things for free because of this arms race competition, I'm pretty uh happy with the way that that kind of unfolds.

I mean it's deflationary. It's deflationary in the pure sense of the word. And and as a consumer, you you could not want anything more from technology than to deflate the price of goods. If you remember, back early in the AI arms race, there was a there was a leak. There was like an internal document at Google that got leaked that basically said

uh there's no mode in AI.

And that it doesn't like open source is so good and it's getting so much better so quickly that even though the Frontier labs are always two to three months ahead on the state of the art,

open source catches up really fucking fast.

Mm-hmm.

And we've just seen that with R1. R1 is as good as O one. Literally, okay, so O3 is supposed to be the new open AI model, which is definitely better than R1. Um, but R1 is basically one twentieth of the cost of O one, and it's getting the same performance. That's crazy. That's incredible, and that's awesome for everybody involved. What it tells you is that.

So, one thing, it means that the motes that these AI companies are supposed to have are not what we previously imagined. And also it means the moat for Nvidia is probably not what we previously imagined. But what it is, is very good for decentralized AI.

It's very good for open source, very good for homebrew. And this is the moat that this is the part of the pool that we play in. You know, on the crypto side, we are all about the democratization of technology and what that can happen when that goes in the hands of individual users and entrepreneurs. Um, a world where open AI owns all the intelligence or Google owns all the intelligence is a world that sucks for us. There's nothing for us to do because we've got all the crappy models, you know, sitting here in the reject pile. Um, that's not the world we're living in right now. So that is something I'm, as an investor and as a crypto person, very, very excited about.

Haseeb, you said this is uh something that the markets got wrong with the response to the NVIDIA price tanking 17% uh on the back of DeepSeek. Uh, we did a podcast with this individual, uh Jeffrey Emmanuel, who wrote this short case for NVIDIA stock. He released this on Friday. Uh, and we when we did the podcast with him, he talked about how this article, which it's a 60-minute read, it's very dense, it's very technical. This guy clearly knows his stuff. Uh, he was telling me about the Google analytics for this article, where it started off just, you know, 50, 100 people reading it as soon as he published it. But over the weekend, it just ballooned to 2.5 million views, starting in the New York, uh, Manhattan area, and then quickly moving to Silicon Valley. And I can't remember the name of the town that NVIDIA is in, but also a lot of uh right in the right where NVIDIA headquarters are. And so he is actually taking the credit for the market movement because it happened, he wrote it on Friday and the market processed it over the weekend, and then it jumped on Monday.

By the way, I think it's correct by the way. You think you've got to be able to do that. I saw this over the weekend. I think it was probably the Spark that started it.

Yes.

I don't know that it wouldn't have happened anyway at some point when Wall Street finally ingested what this meant.

Mm-hmm.

Uh because clearly it has implications for NVIDIA, right? Right. If the AI movement is all happening within the US.

Then NVIDIA is good. NVIDIA's in great shape. If people are not open sourcing

top-tier models that are more efficient than OpenAI models, then okay, that's great for NVIDIA. NVIDIA wants everything to be closed and wants everything to be sold to commercial, you know, commercial enterprises that are going to be charging huge markups on their, on their models.

This is not the world that we're living in now.

And while she would eventually figure that out, I think that being said,

I agree that NVIDIA's stock should have gotten hurt.

Mm-hmm.

And obviously it did, right? But also the market realized that overreacted. So it went down 70%

on Monday, and then it went back up 9% the next day, which was not only the biggest single market cap loss for any company ever in a single day,

but then it was one of the biggest market cap gains for a company on any given day because it just rebounded so much the next day. Um now it's still down from the high quite a bit, but

I think the the the lesson of the story is that

uh it dragged down a lot of other stuff with it, which is what what I think was incorrect, right? Taking what the

what the prospects for NVIDIA and basically their moat in in hardware and saying, okay, this should affect how you think generally about risk assets, that is what I think is a mistake. Because I think it's actually good for all of the risk assets that AI is cheap. I think it's great for all of us. We actually don't want a world where AI is expensive.

Right. Totally. All right. Let's get into the crypto native markets. Let's talk about Bitcoin. Bitcoin started the week at $101,000.5. And then it's up three point uh three percent on the week to where it is now at $105,000 ether. Uh not doing as great. Started the week at $3,200, ending the week at $3,250. Uh, and then of course, I think something to pay attention to is the uh Bitcoin Ether ratio. I don't know how closely you pay attention to the ratio, Hasi, but it did very briefly go below the 0.03 number. It is at above the uh 0.03 right now, almost at 0.031, but it did hit 0.02999, which uh is where Udi Wertheimer and the famous breaking ETH break ETH BTC breaking down below a 0.03 watch party. Uh that that was that that famous level. Uh give me your sense of the crypto markets right now. Bitcoin's still pretty high, Ether doing the same thing it's been doing for two years. Uh just give me your sense on the markets, Sitsib.

Yeah, I mean there was a big sell-off uh triggered by the Deep Seek uh fears over the weekend, and it doesn't feel like alts have uh really recovered. It seems like mostly everything's just been bleeding for the last uh month. So the only thing that really has had any kind of signs of life are Solana. Even Solana is way off the high. So, you know, it hit uh you know, the 280s right after um Trump uh right after Trump Trump token launched, Trump and Melania. Um, and since then it's just been, you know, it's it's it's done the best of all alts, but most alts across the board have just have just bled and haven't really seemed to be able to find their footing with respect to what there is to look forward to in this market.

Did you see this tweet from Ben Cowan? I thought this was super interesting. He goes, the reason why the vibe is so off despite $100,000 Bitcoin is because the advanced decline index, ADI, of the top 100 cryptocurrencies has been in a downtrend since 2021. Now, the advanced decline index, that's not really a metric I see cited too often, but I think the general vibe is of the top 100 coins, uh, most of them are in a just a downtrend for the last two years. Uh and so there are some that are exceptions, like you said, Solana, Dogecoin is an exception, Ripple is an exception, Bitcoin is an exception. But other than that, everything is mostly on a pretty solid downtrend for the past two years. And that's why Ben Cowan is citing the vibe just feels off. Vibes feel bad. So I'm calling this the middle of the market, uh, which is, you know, these are startups. These are startups that you and I invest in. This is uh people's jobs, this is you know the BD in the industry. And the middle of the market has not really felt the same price appreciation that some of the outliers have, like Bitcoin, Solana, um, all these other ones. What's your take on this?

Yeah, there is definitely a bit of a vibe session going on right now. Yeah. Where like, look, if you look in absolute terms, right, the biggest crypto assets, which are uh Ethereum, oh, sorry, obviously not Ethereum, but the has been bleeding. But uh, if you look at Bitcoin, Ripple, Solana, BNB, those are actually doing fairly well.

Um, and then the the big middle of the market is just really hurting.