
Authors: Jonah and Avi | Date: October 2023
This summary unpacks the shifting sands of tech capex, the AI-driven "SAS apocalypse," and the current crypto market dynamics, offering a framework for investors and builders to navigate these turbulent waters. It highlights where real value is being created and destroyed across both traditional and decentralized tech.
“If you are just a mid-stream company buying assets is the stupidest thing you can do and it feels like a honey trap that everybody just falls for.”
“Basically what you're going to see is the revenues that were going to those companies are going to be absorbed into savings from Google and we're going to see the profit go up.”
“The core value proposition of crypto... It is a tool for gambling with your friends and making a ton of money with your friends and being part of a community that just minted millions of dollars for god knows what reason.”
Podcast Link: Click here to listen

Today's episode is brought to you by Kraken Pro. You'll hear more about them later in today's episode. As always, investments in blockchain technology involve risk terms and conditions apply.
Yo, what is going on Jonah? How we doing today?
We're great. I realize that X started this new thing where they don't immediately post the live, so we just have to wait for all of our people to come in.
Oh, just lost Avi AV issues. We're just going to rename this pod from the ThousandX podcast to the audio video connectivity issues discussion forum. Austin, you want to join me for a second while we wait for Avi?
Yeah, let's do it.
While we're waiting for Avi to get back, just figured I'd share that this sort of warm natural suntanned glow that you see on my video feed here is actually not my real skin tone. I'm using a camera filter because it makes me look more handsome.
Yeah, you're looking swave today.
Thanks, man. I live in the town of the Kardashians and the fakeness.
Aby's back. Okay, that was very I always will blame Austin. It's Austin's fault.
Basically, I was just telling Austin we're renaming this from the ThousandX Pod to the Audio Video Connectivity Issues Podcast.
I think that's very fair. We've got 70 people on. We're gonna get Jonah is looks maxing. Oh, yeah. Another thing I shared.
So, basically, you see this warm suntan skin tone of mine right now?
Yeah. It's not real. I'm using a video filter.
Oh, you're using a filter. This is all me, baby. This is all you're going to get.
No filters here. You're naturally tan. I'm naturally like pink when I'm cold and see-through when I'm not cold.
So, that's the downside to being 100% Ashkenazi Jew.
Yeah, exactly. Actually, I got it backwards. I'm see-through when I'm cold and pink when I'm warm. That's how I roll.
All right. So, now that these are all the real ones, everyone on this podcast right now, I want you to know that you're an absolute real one. And so we're going to tell you something only you guys that we're not going to tell anyone else after I post this tweet and that do you guys remember the terminal that we were working on?
I remember you Jonah you remember the terminal that we're working on?
Well, we're pushing out in the next two to three weeks a major major major update and it's going to be I think a very very very competitive product. Uh and the token is currently trading at 800k market cap. That's just for you guys. Not for once I send out this tweet.
Oh, those are the fake ones. Those are the ones that aren't with us from the start. Yeah, they're just the real ones. You they they they're not our friends. You guys are our friends. You guys are the best.
So, how do we get the word out that we're podcasting?
Well, I just have to make this post here. Okay. Be like, we are live. What is the most clickbait thing that I could possibly post right now?
Jonah, Bitcoin's a binary trade. We are live a million or zero. Will bit No, but I feel like it can't be like overtly clickbait. It has to be it can't be like Bitcoin's going to zero. Who's clicking on that?
Yeah. So, I would just say the most clickbaity thing would be like the New York Post kind of has this down. Hot chicks that die unnatural premature deaths. It's usually like the most clickbaity thing. So if you want to post like a picture of somebody with huge boobs and then say woman woman accused of dot dot dot find out now in the thousandx podcast that that I think that would attract some viewership. Maybe not what we're looking for though.
Another way to another clickbay way to do it is to say we're not like legitimate debate about the directionality of markets. Like we're this is uncertain times like we hash it out we disagree. I don't know.
All right. Well, it's out. What I did is I just made a list of all the topics that we're talking about. Let's let's see if that works. Let me retweet one second.
Yeah, just hit that hit that retweet button. Uh where am I? You know, if you were a real one that has been watching the podcast for some time, you would know exactly where I am because this background should be familiar to the people that watch that watch this podcast two, three years ago.
Is it your house in Hyereabad?
What? Why did you of all places to pick?
Because you're Indian. that that is that the only Indian city that you know?
No, I just came up with like the outsourced IT center city just because I was trying to make a joke but it didn't really land. So, I'm sorry.
Awesome. I'm north Indian man. I don't know where I is. It's in the south. You're from Kashmir? From the semi-autonomous mountain.
If I was from Kashmir, I'd be white. All those people up there are white.
Weird. I didn't know that. I know. It's crazy.
Okay, so we're at we're at we're at 200 viewers right now. I'm gonna wait until we hit like 300 and then we can and then we can really kick it off. I'm kind of liking this this this new form of podcasting, Jonah, where we just cold sit here. We sit we cold open. We sit here. We wait for everyone to come online and then once once we get like critical mass, we just start we start going for it.
Let's talk about your house in Puerto Rico for a second. What's out What's out there? Is there a pool?
So that's there's there is there is no pool unfortunately. That is just the patio. It's a nice patio. I just had it I had it repainted about a year ago. This nice blue.
Beautiful. Yeah. No, it's it's really nice. But I have a shelf too. Nice copy of Talmud you got going on there. Whatever that is. It's too small.
No, it's not the Talmood. These are these are all the books from my There's one shelf. It's just like all the books from my childhood. So I've got the Harry Potter series. I've got the Artemis Fowl series. I've got the Alchemist for the real ones that know that. And then I've got like this this book. This book you would love, Jonah. Hold on. Check this out. These these are probably these are probably two of your favorite books in the world. Thousandx Book Club is one of my favorite segments on the show. These are these are probably the best books in the world.
I love both of those. I knew I knew you would love this. I mean, you were the one that told me to get this. I'm pretty sure you also told me to get this one, but I did. Yeah. The Bible and the secret history of the world's best companies. This is a phenomenal book. This is just it's like a riveting piece of text. It really is. It's incredible.
And the the little table at the back that shows VTOL's earnings, I can verify that. Or maybe it's at the front. I forget which. It's either at the front or the very last page. Those earnings are real. I can verify. I was up in there and those numbers check out. So, the net the net profits are pretty pretty insane, which they published a book just before the wild ones started.
Is that Glen Core had a lot of years where they just lost money?
Yeah. That's because they decided basically there's the old adage, if it flies, or floats, you're supposed to rent, not buy. uh obviously that's foulmouthed and I would never I would never coin that phrase myself but an additional sort of correlary to that theorem in the in the world of trading is that if it's a hard asset you're also supposed to rent not buy those things depreciate so what Glenor did wrong was unlike VLO which is asset light VTO rents everything they have the world's largest navy but it's all rented second largest sorry uh behind the United States it's all rented boats right they don't own any any ships for obvious reasons Glenor went the opposite osite direction and went super asset heavy and bought a bunch of mines and assets and cargos and all sorts of crazy things and basically that that emulated their returns to the to the shareholders and this is what happens when you go from being a private partnership to a public company.
So everybody in Glenor like cashed out around 2011 or 2012 or whenever the IPO was. Minted a bunch of billionaires and sent to millionaires and then the company just went like asset heavy and did the whole quarterly earnings thing. Like basically the the best model for these companies is asset light um and majority ownership is employee not outside shareholder. Then Glen Core flipped that and that's that's what happened there. Does that make sense?
Yeah, that that makes that makes a ton of sense. That actually it's a good two things here. One, I'm pretty sure Twitter's down, which is why we have like zero viewers right now. I don't know if Twitter's working on your end, but it just totally crapped itself on my end.
Twitter's down all morning. Yeah, Twitter was down all morning. I think it's I think it's back down now. So, we're because they were bragging about how they're like Facebook has 150,000 employees and we have 30. It's like, yeah, this sort of Thanks, Twitter. Good job.
This is this is actually it's a good segue and I'm curious your take on this because I wonder if there's a correlary between what's happening in the capex spending of these massive tech companies and what you just said which is it's better to rent than to own. All of these companies are saying hey we're going to go build these massive data centers and we're going to spend a lot of money because we want to own the data centers because we think it's going to be smart to own the data centers. There's also the other argument which is that they should not be spending all of this money building out data centers. Uh they should be basically going out to contractors that will own the data centers themselves and rent them just in case at some point in the future these data centers massively deprecate. Uh it's much better it's much better to rent than it is than it is to buy, right? Than it is to actually own them. And so one of the one of the things that I think a lot of people are confused about right now is like are they actually going to follow through on this? Is it better for them to rent? Are they going to walk back their capex spending? Which would be very very very good for their stock prices because it means that stock buybacks could return, but it would be very bad for all of the stocks that have done very well recently in the in you know in the in the in the data center world. So this is this is kind of a middling a middling ground here. We're not really sure what's happening, but I'm curious your take.
Yeah. No, it's a it's a great point you bring up and it's sort of nuance. So at the end of the day, every company is a trading company. Google Google kind of trades ads and compute. Um Glen Core trades physical commodities. The the time when it makes sense to own rather than rent, it's like a real clear distinction. If you are an intermediary in a market, actually let me take an even further step back. Let's use the oil industry as a an analog because it's a hundred years old per the prize and uh the book you just held up and there's a real long precedent there.
So there are sort of three components of any industry in oil. You have production that's called upstream. Then you have trading and intermediation the middlemen that's called midstream. And then you have uh like sales and distribution that's called uh downstream in oil. That's like refining and then taking the refined products and like you know selling them at the rack to trucks at the pump to drivers and you know jet fuel tanks at airports and stuff. So basically upstream, midstream, downstream.
If you are just a mid-stream company buying assets is the stupidest thing you can do and it feels like a honey trap that everybody just falls for. If you are an upstream company, you cannot rent, right? Like if you're if you're digging stuff out of the ground or if you're making making things like you have to own factory, you can't like rent a fac I guess you could kind of go with a contract manufacturer or like own royalties from streams of oil. Like it happens, but the the big big mega businesses own. Downstream you also have to own, right? So where and then we're going to take this analogy back into tech and then try to draw some trading conclusions.
So downstream, you know, you got to own a refinery. You got to own you got to own the means of production and sales and distribution. Otherwise, you're just kind of going drop shipping only gets you so far, right? You you have to own those channels. So basically where mid-stream companies get tripped up is when they try to to do when they try to go from being like a a Glenor to an Exxon Mobile, which makes sense. you want to expand horizontally into the whole market and become this vertically integrated behemoth. If you don't go all the way, um you just end up an asset heavy mid-stream trader company and that's that's a terrible business. Um the companies that succeed like BP and Shell, they did they you know they bought the assets and they followed through.
So going to data centers, it's a very interesting problem. you have like data data center companies like Coreweave I think are kind of a bad bet long run because uh for a variety of reasons depreciation being the number one factor and also um I don't know maybe maybe not let's we could talk through it but if you're Google Google is kind of like the Exxon meta meta and Google Amazon Microsoft these are like the Exxon mobiles of the compute world and so given that they are vertically integrated I think it does make sense for them to own the assets and the reason why it makes sense for a vertically integrated company to own assets where it does whereas it does not make sense for a mid-stream company to own assets is because they extract the full optionality out of those assets because they can pull from their supply during times of shortage they can push into their into their sources of demand that they own during during times of excess and so like those assets actually produce more option ity which justifies them if you're just an asset owner like Cororeweave there they're going to be real cycles for that business like a mid-stream asset owner so I think basically yeah I agree said Cororeweave is basically a landlord for chips and that their assets depreciate rapidly that is that would be a scary business for to be in just like amassing amassing physical piles of chips and wires and just like sitting on them. Um because I assume I mean the the the only the only the only major difference might be are we have we reached the pinnacle of chip development like is is is is chip development going to slow down a ton because this is this is this is really this is actually an interesting question because it's kind of what happened in the Bitcoin mining space where Bitcoin miners were like horrific investments uh early on because the rate of development was super super super super super fast right you were constantly like every year they're coming out with new miners and then suddenly the pace slows down and it slows down and it slows down and actually now the development timelines for Bitcoin miners are are pretty long because we've reached like literal physical constraints on what you can do with these chips.
Yeah. Uh and so it's like minor difficulty is just skyrocketing. It's it's skyrocketing and it's not it's not because of uh chip development and like faster chips. it's because people just keep amassing more like that we're producing more miners, right? Whereas in the past the difficulty was skyrocketing because a new chip would come out and like 20x the efficiency of of a Bitcoin miner. And so the question is are are we does does it become a better business over time as chips become more stagnant because we've reached physical constraints?
You're actually the chip guy now. So yeah, I am. I mean, I haven't really talked about it much. I'll still keep it on the DL. But in general, like the um the way that I view it is like amassing amassing piles of hardware in data centers is going to converge towards the returns of like a real estate landlord business, like a commercial real estate empire, which I guess is good business, right? Like they're not going to go bankrupt. Glen Core didn't either. But it's not going to be like the hyper growth exciting thing to do because again it's a you're a mid-stream company and you're just you're just a hard asset owner. You're not like rent you're not like nimly renting the assets you need to like use high leverage to bootstrap some to to basically to surf some mega trend to surf some tsunami wave. Um, kind of like Bitcoin miners weren't great business until they pivoted to AI. Like maybe maybe there will be a next big thing that Coreweave can latch on to after AI, but somehow I don't think so. I think it it I think this is the final wave in the set to stick with my surfing analogy. And once it's once they've written that out, it's going to it's going to go down. I I wouldn't be long coreweave here, that's for sure.
Um, however, I don't think it's a bad idea for Microsoft and Google and the other hyperscalers to be amassing physical infrastructure because they're actually going to need it for a long time for inference and there's tremendous optionality in there. Like let's say that we go into a glut compute, they can always there's never going to be a shortage of demand from their internal needs, right? Meta advertising and just usage. Google will always have a place to, you know, put the compute. This is why Exxon has assets, right? Like if when there's a glut of oil, they can just shove it into their refineries and tell their refineries, "You can't buy from anybody else." Same thing with Google owning a data center. If the external demand for compute dies down, they'll still have internal demand for compute that far outstrips their physical infrastructure ability to provide it. So it makes sense for them to own. Um that you kind of getting what I'm
Yeah. No, I get I get I get what you're saying is that they can they can pick up the I also think a big a big part of it is that they just have the they just have the money Yeah. to basically eat eat the low times like eat the lean times. But I do think that the amount of money that they're spending is sort of making a bet on exponential growth of demand. And if we don't see that exponential growth for an extended period of time, I think that we could we could see some issues. But I I'm of the opinion that it's possible that we're actually going to see some walkbacks in in the capback spending that these these companies are doing. Basically, everyone's gotten so nervous. I mean, Google Google's now trading Google's now trading at 300. Meta Meta has been doing terribly. Um, they sort of been have been caught up in this in this route of tech companies across the board. But I I don't think that's I don't think that's very fair. I think that we're this is actually like this is a pretty good time to go buy Google now. This is a pretty good time to go buy Meta. It's a pretty good time to go buy all these all these uh mega cap tech companies. Uh because I do think that one thing that we're not quite appreciating is just how horrible the SAS apocalypse has been. Uh and but what it what it also means for for the large cap companies, right? Basically, everyone that pays insane amounts of money to go use Atlassian, to go use in it, to go use Workday, like the amount of money that is spent on Slack and Adobe and Salesforce, like the companies that can cut these people out now because of AI, the the speed of AI development and what you can build internally, they're going to save there's going to be tremendous savings. like the revenues from all these companies are basically going to get back in and reinvested in the mega caps that can actually create them internally. And so this is like the best trade of all time right right now. I mean, it's already sort of played out. So I don't know if you want to take I don't know if you want to take the short side here. Like I don't know if you want to go continue to short Adobe down 25% continue to short Atlassian down 47% year-to- date, right? But basically what you're going to see is the revenues that were going to those companies are going to be absorbed into savings from Google and we're going to see the profit go up. And this is this is what people talk about when they say AI is going to improve productivity and AI is going to streamline companies and it's going to you know basically increase margins. It's all this. all the tens of millions, hundreds of millions of dollars that are spent on these companies that produce billions of dollars of revenue in total that is going to collapse now because I think a lot of these things are going to be built in house. Like I wouldn't be like, you know, Salesforce is is is a great example of this. Uh in the next three years, I guarantee you that a lot of people are getting rid of Salesforce because they've just built their own internal tools. And so this is like the SAS apocalypse to me, what everyone's talking about and all these shorts. I think is is really relegated to the BTOC area. It's really relegated to these large companies that can afford to that can afford to develop in-house. Um but BTOC I think is a little bit more safe just because the average consumer is really dumb. Uh and and it's it's it's actually like once they have buy into a product, it's quite difficult for them to switch. And distribution, distribution is still key. Like when it comes to uh when it comes to the consumer, the consumer is never going to make the calculation. Could I build this myself or should I spend money to go pay for this thing? Businesses are make that calculation every single time they sign up for a product. Should I build this in house or should I buy this from somebody right off the shelf or this is exactly the discussion that we were having. Do you build it yourself? Do you own it or do you rent it? And for the most part, because there's been such a high bar for entry for building software and so expensive, the answer is rent. You want to rent because you don't you don't want to, you know, it's inefficient to build it yourself. But now I think it's very efficient to build it yourself. So that's sort of I'm still I'm still on this I'm I'm still on this trade. I think, you know, you you you buy you buy you buy the mega caps, you short everything else. You short you short the service providers to the mega caps right now because there's they're just going to figure out how to do everything themselves at this point. I think that's actually probably a pretty actionable trade.
The Salesforce stock price, I mean, like you got to be careful though. Like the Salesforce stock price is almost down to the 2023 lows. It's trading um 189 right now. All-time highs was the beginning of 2025. It was 362 uh when Trump got elected, the day after. Now it's down like 50%.
The the way the way that you short generally is you wait for the bubble collapse to occur, but then you you don't short on the way down, right? What you do is you wait at some point there's going to be a violent bounce. But if you believe strongly that you're in a secular downtrend, then that's your entry, right? You wait for Sal for Salesforce is almost at the 2020. It's actually not almost at the 2022 lows. The 2022 lows are 131. It's currently at 190. But basically what you wait for is like a one to two week period where the performance is we're up like 15 to 20% on these companies and then you go short. Right? That's that's the way that you man you manage a short is you short you want to short these things in in strength. And so if if I'm constructing a trade here, I'm I'm definitely like I'm still a buyer of Google. I'm still a buyer of Amazon. I'm a buyer of Microsoft. Um I'm not telling you to short Salesforce and Atlassian here, but I do think I do think this is a trend that's going to that's going to continue for an extended period of time. I mean, this is like I view this as the alt market blow up equivalent for crypto, right?
Yeah. like when when you hit a bare market, the that is useless goes to zero. And I think we're going to the thing is it's easier actually to short Salesforce and Atlassian and into it. It's easier to short them because they actually have revenues that you can look at. So all you have to do is say, okay, let's see are their revenues actually going down? Are people actually canceling services? If you don't think that there's anything to reverse that trend, these are phenomenal shorts.
I mean, I like your framework for shorting stuff. You know, at the beginning of the year, you were talking about shorting meme coins because they just bounced. And looking at my least favorite one ever, whiff, or my favorite one to to hate, the one I love to hate. It traded from 25 cents up to 50 cents and now it's trading 23 cents. Like in these violent bare markets, you get squeezes. If you're just sitting there ready to with the hammer in your hand, ready to play some whack-a-ole, you can make some money. You get some you can whack some moles.
Yeah, I mean it's it's it's really the same trade as as as the as the crypto trade.
How is the crypto trade going by the way? There's a cryptoorm on brand wrote when crypto I think I mean I sound like a broken record. We I think you just got to wait it out. It's not the hot item right now.
No, I mean our daddy Mike Epalito tweeted something. Uh try to pull it up here. He's saying um he expects crypto to go into a like a nine month minimum bare market. Let's see here. I'm going to read his post because I think it's relevant and I want to talk about it with you. I'm kind of wondering the same thing. Um some thoughts on this market. I'm gonna try to do my Kiple's voice. Some thoughts on this market. I think it's likely we're entering a full-on crypto winter. I'm also open to the idea that this bear will be as bad as 2022, perhaps even as bad as 2019. The short-term reason for this is that their industry is in an air gap created by unsustainable valuations and regulation. We've been pounding that table on this podcast for years. Historically, valuations in crypto have been driven by the hot ball of money. Money comes in and because it was literally illegal to generate value for tokens, revenue and cash flows were entirely disregarded. Prices were set by the amount of capital times the supply of tokens. Uh the sexier and more risk on it was the higher it went. There are two things that are different this time. The biggest difference is that it's clear that there will be a regulated path for crypto projects. This is I'm skipping ahead. This is good, but it pres prevents an presents an obvious problem for protocols valued purely based on speculation. Once there is a regulated way to generate cash flows and not get thrown in jail, that is all the market will want. So, what is confusing many investors and founders right now is that fundamentals are growing, but tokens are still selling off. This is because we're resetting how valuations will work and the starting price for almost every project was way too high. We said that on this podcast. Additionally, crypto is getting absolutely mogged by AI. The last couple years of memecoin stupidity are catching up with us and unfortunately, we didn't build anything useful. All in all, my best guess is it is a 9 to 18month bare market before things get better. Oh my effing god. Avi, what do you think?
I think that's pretty I think that's pretty reasonable, but it sort of depends on what you think of when you say bare market, right? It's it's there will be there will obviously, in my opinion, be a bare market for really shitty, terrible assets. Uh, but you have things like Morpho and things like Uniswap that are getting bought up by large institutions right now. You have things like Hyperlid that are generating real capital. Um, I'm crossing my fingers that one day syrup will stop sucking ass and actually start going up again. But, you know, the these these things are actually generating real revenues and becoming real companies. And we've talked about this at Nauseium in the past, and I don't really want to beat a dead horse, but this is this is the dot implosion moment where your pets.com it goes to zero, but everything else ends up going up, right? like all the stuff that is actually a good company run by smart people that are looking to exist in five years where most of these crypto projects are just looking for a quick buck. The things that actually are trying to build I think are are are very very very good trades because they're right now what's happening is they're they're being dragged down by the broader market. They're being dragged down by this idea of of a bare market, right? I mean like Hype's Hype's trading as much volume as Coinbase now. That's insane. Hype's not valued nearly where Coinbase is. No coin. Although it's they're converging quick. Holy moly. Coinbase is is just imploding.
You know what? You're you're right. Like this really is the technology market in 2001. Pets.com is or I guess maybe in 2000. Pets.com is dying and Amazon is getting hit in sympathy. But, you know, the next 25 years are pretty bright for the projects that actually build. Like another one that I I didn't want to leave out, my favorite aerodrome, or as my Italian and French friends pronounce it, Herod drrome. Um, I think that I think it's got a lot of upside from here. It's just the question is like when does the good stuff stop getting dragged down by the bad? Mike Epalito says 9 to 18 months. Um, to me that just feels like, and I don't blame him because I pull stuff out of my ass all the time. Uh, that feels like a time frame that's been pulled out of his ass. Now, I don't I don't have any issue with that. I just it's I can't help but wonder and we should probably discuss why. Like, what is that? In the past in the past, there were like catalysts. It was like, okay, well, Luna and FTX just blew up and volumes are down 90%. people probably won't re-engage for at least a few months, maybe years. Here it's like what all the back the regulatory and and narrative backdrop is so constructive. It's like why why should it take 18 months for the valuation convergence to occur? There are huge trades here. Short garbage against being long good stuff. short Coinbase, long hyperlid, short whiff, long uh long long aerodrome. Like what am I missing? Why does it take 18 months for the market to correct? Or is Mike getto right?
I think I think what what happens here? Well, there's sort of two things that we need to we need to talk about. If you're if you're an investor and you're just thinking about like how do I how do I think about trading the crypto market? Well, two things are true. One is that we're having a massive blowup moment and that this has been talked about and that it's probably going to take some time to wash out all of the exuberance that we experienced. I think we pulled forward a tremendous amount of value just because of what Trump did pumping up the crypto markets, launching his Trumpcoin, launching the grifty ass Melaniacoin, you know, basically trying to bring in like right Ba ba b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b b