저자: Kevin · 외국인
요약 보기
All-In 팟캐스트(2026.3.13, 이란 전쟁 유가 충격·AI 매출 주제)의 오류를 짚는 영상. IEA 비축유 4억 배럴 방출은 발표 전 이미 시장에 반영됐고 유가 단기 변동성 위주라 전쟁이 단기간에 끝날 것을 시사한다고 봄. 참석자들의 유가·중국 원유·AI 매출(Anthropic 추정치)·AWS 신뢰도·데이터센터 비용·부유세 계산 등 수치 오류를 지적. 추가로 이스라엘 요격미사일 부족, 상륙부대 중동 이동 등 확대 신호를 언급하며 전쟁 장기화 가능성도 제기.
다듬은 글
In this video, we're going to break down the inaccuracies and mistakes the All-In podcast crew make on their March 13th, 2026 podcast titled Iran war oil shock off-ramps and AI's revenue explosion. Some of these are just glaring errors and mistakes that we have to analyze. Some of the things they say are useful, though also lack context.
So, in this video, my goal is not only to provide you more perspective, but save you time. So, let's get started. First, we're going to go 8 minutes into the podcast. Donald Trump says that that the war would be over very soon, says Chamath on Monday. And Chamath says that because Donald Trump says that oil prices fell from $120 to $90 per barrel.
This was a victory lap by Chamath for Donald Trump, which lately the podcast has been doing a lot of victory lapping for Donald Trump. Except, unfortunately, Chamath is entirely wrong here because for those of us who were actually paying attention on Sunday, we recognized that on that same evening where oil prices skyrocket to $120 per barrel.
By the time we woke up in America, we heard discussions of well, guess what? IEA releasing 400 million barrels of oil from the global strategic reserve, which is a combination of 32 different countries. And those discussions are actually, in my opinion, what heavily helped tank oil prices to $90 per barrel.
Because just 2 days earlier on last well, last Saturday, so a week ago, Donald Trump actually suggested that we might use special forces to extract the highly enriched uranium 460-ish kilos of highly enriched uranium out of Iran. That was an escalatory comment, which maybe would offset this Donald Trump saying war will be over very soon, but let's be clear, releasing 400 million barrels of oil from the global strategic reserve would help offset probably somewhere around 20 to 30 days worth of the capacity shut-ins that we're not getting out of the Middle East right now.
Obviously, because other areas are still producing oil, Russia, India, United States, not everything is damaged here. So, we're really just trying to replace or offset what we're not producing at the moment. So, in my opinion, these Trump comments, sure, they don't hurt, but they're kind of balanced by some of the other things that Donald Trump says, which are escalatory.
To me, this was just straight shilling for Trump. And while Chamath does later talk about, very briefly, the IEA's release, the International Energy Agency's release of 400 million barrels of oil, he kind of implies that that came 3 days later, which it's true, the official announcement came 3 days later, but the actual murmurings that it was happening and that the meetings were occurring happened that Monday morning and on Tuesday.
So, the markets already heavily priced it in, which you could see because when the official announcement came, oil didn't actually really move much. If anything, it went up because people were hoping for more. And this volatility in the short end of the oil futures curve is actually really important, which the All-In podcast doesn't mention this difference at all, but all this volatility we're seeing is on the short end.
So, you know, within about 30 to 45 days on the oil futures, we're not really seeing long-term oil price movement, and we're also not seeing a lot of call options on oil above 120, 130 dollars per barrel. This suggests that most analysts today and traders, hedge funds, people making bets, whatever, also agree that this war will probably be on the shorter end.
Which David Sacks does give us some administration insight, it's one of the reasons that, you know, even though I usually think of David Sacks as a warmonger, it's one of the reasons I kind of like this sometimes just getting a quick little feel for what David Sacks' opinion is because I think he gets insights from Trump and then kind of sort of like slips them out on a pod.
And he talks about how Donald Trump is the kind of person who wants to have a quick strategic operations just like operation Midnight Hammer last year where we dropped bunker busters on Fordow and Natanz, the bunkers for uranium enrichment likely full of centrifuges to enrich uranium potentially to weapons grade levels.
That mission was obviously an in and out and David Sachs suggests that Donald Trump has a preference for this sort of in and out. Okay, interesting. And and hopefully he's right, but it doesn't really reconcile with something else the All-In podcast didn't talk about at all, which is the fact that now there's talk about potentially 5,000 Marines and sailors moving to the Middle East for amphibious assault potential purposes.
Now, we talked about this on the channel yesterday that an amphibious assault might be moved into the region so that we can take control of the Strait of Hormuz. Maybe. We will see, but they are now being relocated to that region from places like Japan or the coast of California so that way they could deploy if they need to.
To me, that kind of sounds escalatory. So, as a result, just a quick summary on this, I kind of think you know, Chamath gave Donald Trump a false victory lap here and they completely ignored some of the other things going on in the region. Though, it is also worth noting that the we do have information that the strikes on Kharg Island by the United States purposefully did not strike the oil facilities on the island, which is good because Iran does threaten counterattacks to other ally or, you know, Saudi or whatever UAE oil infrastructure in the region if their infrastructure is hit.
And these strikes on Kharg Island right here, you could see were primarily focused on the airfield here and then locations here in the more mountainous terrain where we have oil facilities are here on the bottom left and over here on the bottom right uh the right side. Now, a lot of these right here are uh oil um storage tankers.
What's very unique about these is they have floating ceilings. So, when you can kind of see a little bit of a shadow here, the less of a shadow you could see, the the less likely uh it is that this is empty. So, the less shadow you see, the fuller it is. I mean, think about it. It's sort of like the shadow in a cup, if you will, right?
Like, if you point a flashlight or the sun from one angle, you're going to see a pretty nice shadow. If that roof is all the way to the ceiling cuz it's full, you're not really going to see any shadow. Kind of unique how that works, but I think it's useful to know this because David Saks does go on to a little bit of a rant about how much infrastructure is potentially getting damaged and the humanitarian costs of this in the region.
Now, there's some mistakes that Saks makes here, and I think they are somewhat important because I do think he's going a little bit too extreme on some of his commentary. Let me explain. Now, David Saks talks about how desalination plants in the region are potentially also getting struck by Iran. And this is dangerous because when the strait reopens, not only are we going to be down in capacity on oil, which oil infrastructure has gotten hit, fortunately not here on Kharg, but who knows that could happen.
Uh what we also end up with is this idea that desalinization desalination, there's always that temptation to put that extra Z in there. Uh desalination plants are getting damaged, and this is a region, the Arabian Peninsula, that relies heavily on water because it's mostly a desert. David Saks says the Arabian Peninsula has a population of around 100 million.
It's actually about 93 million, but I don't think that's that big of a biggie. I mean, the rough approximation is right here. Now, he says that the entire region relies on desal. This is unfortunately objectively wrong. The Saudis, who are really quite rich, they have the largest desal uh setup in the world, but they only represent about 1/3 of the population of the Arabian Peninsula.
And they're also on the richer side and more capable of repairing and rebuilding. Now, who is not? Yemen. Yemen has about 1/3 of the population in the region as well on the Arabian Peninsula, and Yemen doesn't use desal. So, about 1/3 of his desal argument is just blatantly wrong because he doesn't mention that Yemen doesn't use desal cuz they can't freaking afford it.
So, some of the arguments that the David Sachs makes here, I think, are just really trying to like set up the extremeness of Iran trying to cause all this damage in in the region, which is true that is part of their goal, but he kind of misses the point on this one well, by at least a third. Now, talks about destroying this humanitarian infrastructure, important, does sound like he's trying to sound anti-escalation.
He kind of wants this war to end quickly, so do a lot of people, and that's what markets are betting on anyway. Now, to me, I wonder if that's a hint from the Trump administration that maybe it will, but again, it doesn't really reconcile with those amphibious units moving into the region. Who knows though, maybe that's a Trump's strategy to secure the strait, which obviously would be a big positive for oil.
I wish they talked about that on the podcast, but they didn't. David Sachs does talk about Israel having a social media blackout and then completely glosses over what this is for. Now, I'm not a shill here for censorship. I got censored when I ran for governor in California, and that censorship has no place in our society.
But, and I'm also not a shill for Israel, but let's at least be clear here. This is a social media blackout specifically on IDF opsec and military movements, which to some extent is somewhat reasonable given how small Israel is and everybody can get to see what the military is doing in Israel. So, small fact check here, but it's worth mentioning.
David Sachs then kind of goes down the line and says, you know, if this war keeps going on, we're going to end up finding Israel's going to use their nukes. I think a lot of people talk about this and they talk a big talk that Israel's going to start using nukes. I I don't know if this is like uh more sort of warmongering or kind of going back to like this is why we got to crush the Iranians.
So, like some of what Sachs says is anti-escalation, but then some other parts of it are not, and they're actually escalatory. I'm having trouble kind of reconciling some of what Sachs says. But, I can tell you that per Semaphore, Israel is running critically low on interceptors, which is bad because obviously the more damage that Israel ends up experiencing, the worse it is for this war because it will lead almost certainly lead to more attacks from the Israelis or worse, some form of uh you know, ground occupation of parts of Iran.
Iran is absolutely massive and has a massive 80-plus million uh person population, so I don't think the Israelis uh really have a chance to occupy that, if you will, like uh like uh they did in Gaza, just because of the size and population difference, but a ground operations would it just wouldn't surprise me.
Now, about 17 minutes in, Jason, the guy who says he wants to slice me up, not entirely sure if that's just the way he is trying to become friends with me, but Jason told me on the friend on the phone that he wanted to slice me up. Kind of weird, but anyway, I have to actually give Jason credit here because I think he's right when he says that the longer this war goes on, the more likely it is Democrats are going to clean sweep midterms.
First, you have the Epstein files, which Jason downplays, probably because he's in them. Uh then uh you have this war, which is exactly what people did not vote for. No new wars is what people voted for. Uh and then of course, you have the uh Customs Enforcement raids, which have gotten a little extensive, uh and obviously have led to some innocent lives being lost, though Donald Trump is trying to solve that.
Getting rid of Kristi Noem probably helped. But, here Jason does acknowledge that MAGA 2.0 might die uh because the administration is ignoring the working man and ignoring inflation. And in fairness, uh they don't mention this, but in fairness, this war and tariffs increase inflation. Even if it's just a transitory effect, we're tired of hearing transitory because so far transitory has been going on for about 6 years now, and it's turning into a pisser.
Now, with that said, let's move on to the next comment. Brad says, "We can take care of ourselves. We produce about 20 million barrels of oil a day and consume about 20 million." All right, let's clarify that a little bit. We only produce about 13.5 million barrels of crude, but he is also including natural gas, biofuels, etc.
Small clarification here, yes, we're self-sufficient with our consumption, which does include all of those other sources, biofuels, natural gas, whatever. Uh so, to to some extent, he is correct here, just kind of uh sloppy in the way he set that up here. Obviously, uh though when when you have a supply crush or crunch in one area of the world, prices go up, even though we are self-sufficient on oil.
I think his point here is really just to try to de-escalate the war and say, "Why are we getting involved in this?" And obviously, it has to do with the highly enriched uranium that the Iranians have, about 460 kg, uh of which with the um United Nations weapons inspectors did verify Iran has. They did as of just before Operation Midnight Hammer last year.
That's through the IAEA. IAEA, International Atomic Energy Association Administration, something like that. But anyway, uh you know, that's just what we're being sold to. So, we don't actually entirely know if that's the reason or because Donald Trump wants regime change. We know Bibi wants that. Who knows?
But the point here is Brad does sound a little anti-escalatory, which is good. Chamath then, though, makes another oopsie-doopsies. Chamath says that sanctioned oil from Venezuela and Iran make up about 20% of China's oil supply, which covers 100% of China's transport, cars, buses, planes, etc. No, this is wrong.
China's transportation sector uses about 50% or roughly half of the oil that China needs, which means Chamath was off by a factor of two and a half times. And that's if you assume 20% comes from Venezuela and Iran. I think that number's high. Most estimates have it dis-pegged between 10 to 20%. So, it's not even guaranteed that it's 20%.
On top of that, Russia supplies nearly half of Chinese oil and India supplies 38%, which means no, China is not getting whacked the way Chamath wants people to believe because of this shift in oil production in Venezuela and Iran. Instead, if anything, China might actually be benefiting because Donald Trump just lifted sanctions on oil sales from Russia.
So, Russia can now sell oil to the Chinese sanction list or to other countries' sanction list, which is remarkable because the whole point of sanctioning Russia is to not enrich Russia because that just makes our war in Ukraine more expensive, but that's exactly what we're doing here. And of course, no mention at all from the All-In podcast on the potential negative ramifications of some of of Donald Trump's actions.
Uh well, because I think they're just straight chilling for Donald Trump and we're trying to provide a balanced analysis here. Anyway, Chamath, as a result of his, you know, wild numbers, thinks that the meeting between Trump and Xi Xi Jinping is going to end this war. He argues that the Chinese need this and so they're going to convince Donald Trump to exit this war.
I think that's wild to think that the leader of China is going to convince Donald Trump to end this war, but who knows? It's it's an idea. Anyway, then we get into artificial intelligence about 27 minutes in. And on the artificial intelligence side, Brad Gerstner claims that Anthropic has a six had a $6 billion month in February.
I think that is about five times as inflated as what reality is because recent reports have indicated Anthropic has about $19 billion of ARR, which is way less than the $17 billion dollars, sorry, $72 billion dollars a year Brad is suggesting. Remember, if Brad is saying that is 6 billion dollar a month, 6 * 12 is 72 billion dollars.
Now, if current and the latest reports, because they said Anthropic is sitting at about a 14 billion dollar run rate, and then Brad comes in and says 6 billion dollars a month, so there's a mess to clean up here. If we assume 14 billion, then Brad is overestimating Anthropic's revenue by 5x. If we assume the latest facts, which are about a 19 billion ARR, then we are still taking Brad's information overstating the revenue at Anthropic by 3.8 times.
Now, I think this is kind of important because Brad's the guy that's going on here going, "Yeah, I keep putting more money into OpenAI and Anthropic and AI AI AI AI is moon, man." And he's really off on his numbers, which just makes me wonder how much of his success comes from blind luck. But, I I don't know much about his history, so who knows?
That'll be for other people to evaluate. Uh now, the entire crew seems really optimistic about Anthropic and OpenAI going public. Makes me wonder how many private shares they have that they're trying to bail out on and dump onto retail, but whatever, that's me speculating. Let's stick to the facts. Chamath talks about how AWS, Amazon Web Services, is uh well, basically has 12 nines of accuracy.
Let's correct this a little bit. First of all, when somebody says nines of accuracy, what they're talking about is nines after 99%. So, Chamath says AWS is 99.9 and 12 nines accurate. Okay. First of all, Amazon advertises that they are 99 and 11 nines accurate for you not losing a file for data corruption.
That is different from AI accuracy a lot. It's also very different from server uptime. Now, you might think, "Okay, what's the big difference here, Kevin? Whether it's a few nines of accuracy after 99.9% or you know, 11 nines thereafter. What's the big difference?" Well, the big difference is if you used uptime, for example, if you took 99.9% with nine nines of uptime after that, you would only have one single second of AWS downturn downtime every 3,171 years.
In other words, like never. You'd like never have downtime on AWS. Yet, when you actually take 99 and 99.9% three nines of downtime for AWS, you end up with about 8.7 hours of downtime on AWS every single year. So, when Chamath implies that AWS is 12 nines accurate, we don't really know what he's talking about because there's no way in a hell their AI is 12 nines of accurate.
Amazon advertises that they're 11 nines of unlikely to corrupt your files after 99% and if we're talking about server uptime, then we're nowhere near those estimates because we all know AWS goes down. We all hear about it all the time. Now, clarifying another oopsie doopsie from Chamath. I hope that's not the kind of research that he does on his SPACs.
Who knows? Anyway, Brad is super excited about OpenAI and Anthropic's revenue and Chamath comes in a little bit more skeptical. Chamath says token costs are up like 3x, but his revenue from this that he's can generate from these tokens is not up 3x and he thinks the same is true for AI companies that really Anthropic and OpenAI, they're selling picks and shovels in the gold rush to the speculators like Chamath and he thinks a lot of the speculators are going to fail including potentially some of his own companies that he invests in and so he's not seeing those high ROI numbers that Brad is really shilling and maybe be some of this is because both of them aren't really good at actually getting into the nuance of the numbers because they've both made some pretty critical numbers mistakes.
They just seem to really hype things up on the podcast. And so we have to kind of bring real perspective back. Now, David Sacks does say that which I actually thought this was a really good point. He says for years we've had a major engineer shortage and AI's coding tools actually help companies hire more software engineers because software engineers can do more work with AI coding tools.
It's kind of like an engineer that might take 20 hours on a project on one project could take those 20 hours and work on three projects. So you end up with more products. And this is also in part why we've seen some software engineer hiring ramp up. Probably because more companies are developing their own software.
I mean, heck, we have about nine to 10 developers working on our artificial intelligence. You can learn more about that at househack.com what we're building with real estate artificial intelligence and sort of our release time frame for that. In fact, we have an option where you can buy our real estate artificial intelligence with lifetime access, but once the full suite comes out within about the next three months and at the end of Q2, we're going to a monthly model that might be two to $300 a month.
So huge opportunity to buy that. But our developers are are much more productive and so we see it as well that there's a lot of money to be made with new products that people are creating like what we're doing with real estate vision and valuation based artificial intelligence that nobody is doing. Zestimate, Redfin, none of these companies are doing this.
So we're building something that we think will be huge for investors. I mean, imagine going on your phone and being able to see what deal in the entire country would make you the most money if you bought it today. And you could sort all the deals on the market by how much money they're actually going to increase your net worth.
Kind of wild if you think about that. Now, anyway, the point here is Sacks says software engineer hiring is up. And I I I agree with him. I actually also agree with Jason when he says it's really bad that states like New York want to ban artificial intelligence on healthcare and legal work. Mostly because Jason makes a good point.
Startups use legal or or use AI to analyze, hey, should I sign this contract? Or individuals use it for healthcare. What is What is my doctor saying? Is this accurate? It's really a sort of poor man's fact check, if you will, or consultant. And it's actually really good. So once again, I I hate that I have to agree with Mr.
Slice You Up Jason. But he makes some really good points that startups are able to do a lot of damage now with artificial intelligence, and it's a really good point. 44 minutes in, Chamath says that 1 gigawatt in Arizona costs about $50 for artificial intelligence. I do think this is a little bit exaggerated.
Cushman and Wakefield argue that data center per gigawatt costs about $11.7 billion to stand up, and then maybe another doubling that for Nvidia chips if you're just trying to set up a regular server. So here, I'm I'm not trying to say Chamath is wrong. I'm just trying to add context. You can usually get a standard data center up for about $20 per gigawatt, which means Chamath is exaggerating the typical data center cost quite a bit here.
Now, in fairness, if you're talking about bleeding edge, which he doesn't clarify, and you're going to get the latest Blackwell Vera Rubin chips, yes, then the Nvidia chips could cost you all the way up to $50 for a non-traditional all in on the latest hardware AI supercomputer data centers. And again, my goal isn't just to bag on Chamath or these people from the All-In podcast.
It's really just to add perspective to you and try to give you not just an idea of what they're saying in case you got bored watching their stuff, but also give you extra perspective on what they really left out. Now, let's also touch briefly on this next item. About an hour in, the David Sachs says Vitalik donated dogecoin.
This is not true. He donated shib, 100 million of shib to Future Life Institute FLI, the Future Life Institute FLI. And then in 2021, he donated also to India's COVID relief funds with shib, not doge. Small clarification, those in crypto might care about that. Anyway, about an hour and four-ish minutes in, Chamath says data center cancellation rates are rising in part due to protest.
This is actually I'll give Chamath credit here. This is actually a really interesting point here. So, they put up this chart right here. It says people are rejecting AI data center as a data centers as electricity costs rise. And so, we've had a lot of, you know, here somewhere around seven estimated gigawatts of cancellations coming in 2026.
Chamath says this is potentially due to protest. And the Guardian actually has a really good point on this or article on this. This is a summary from this. Chamath says about 40% of data centers are getting canceled, but you could actually say that about 2/3 of targeted data centers have been halted.
About 98 billion dollars of projects have been delayed or canceled, and a lot of it potentially due to protest. So, Chamath in fairness to Chamath, actually makes a really interesting point here about how that could affect the economy. Now, David Sacks makes another oopsie-dupsie, and this one's a pretty sad math error.
Everybody should just roll their eyes at this one. Now, I want to be really clear here. I am anti a wealth tax because that is a form of double taxation. They tax your income, then they tax you on your wealth. That's sort of like coming back for a second dip. Kind of wrong. An hour and 13 minutes though, David Sacks says that there is a national proposal by Bernie Sanders to tax wealth at 5% per year over 20 years.
And after 20 years, you would therefore be left with zero. So, basically, if they tax you 5% every year, he just says, "Well, 5 * 20, you'll have nothing left." Okay, this is still robbery, but David Sacks is totally wrong in the math because every year they tax your wealth 5% you're left with less to tax the next year.
So, the actual correct formula is 100 * .95 to the power of 20. Remember you do exponents first. This means after 20 years you'd actually be left with 35.8% of your wealth. Which is still a robbery of 64.2% because it's a double tax, right? Still messed up, but this was a rudimentary math mistake by David Sacks.
I I'm sure he like I think he's just trying to again exaggerate that Democratic policies are really bad and I agree that this one's bad. They talk about you know, the former CEO of Starbucks left to Miami because of the wealth tax that was passed in Washington. Totally fair. I'm anti those sort of taxes as well and I agree that Newsom is probably going to end up positioning for a national wealth tax as well, which you know, is a Democratic talking point, but I do think David Sacks is being unfair to the argument that is being made by taking the extreme of Bernie Sanders and then doing the math wrong on top of that.
So, there are some mistakes in this podcast. Overall, some of it was interesting. I like the comment that Donald Trump will hopefully go for a shorter term war, but there were a lot of mistakes in this that I think were careless and I think the All-In podcast could do better. Uh if you like this format, please leave me a comment and I'll do more.
In fact, you can even leave me a link of podcast you'd like me to do a breakdown like this on. Thanks lso, check out meetkevin.com if you want to check out my programs and get lifetime access to those. We do have a coupon expiring on Wednesday. Thanks so much. .
Comments 0