# The Dirty AI lie : How the GREATEST bet in human history started to crack in June 2026?

https://www.youtube.com/watch?v=WcckBmkauBQ

[00:02] How do you think about this bubble talk that has been going on for the last few months especially?
[00:09] I mean, I think it's quite possible.
[00:12] Ladies and gentlemen, on 25th of June 2026, Apple did something that it has never done in its history.
[00:16] In the middle of the year for no new product, it just raised prices.
[00:22] MacBook Air is up by 18%, iPad Pro is up by 20% and Apple TV is up by 54%.
[00:33] Apple said yesterday it is immediately raising prices on the products.
[00:36] The company says soaring memory chip prices are driving up costs and the AI boom is a major factor behind the surge.
[00:43] The AI trade is leading to real near-term inflation.
[00:48] And when asked why, Apple said something remarkable.
[00:50] They said, "We have never seen a competent price increase this much this quickly."
[00:57] And the reason your laptop got more expensive is because of a war being fought over tiny memory.
[01:02] Chips thousands of kilometers away.
[01:04] And look at this graph.
[01:07] In 2020, before Chad GBD existed, the four biggest US tech companies spent combined $90 billion on capeex.
[01:13] In 2023, they spent $147 billion.
[01:16] In 2025 that number went up to $410 billion and then in 2026 it is up to $725 billion.
[01:25] So in 6 years the capex has grown 8x and all of this is coming just from Amazon, Meta, Google and Microsoft.
[01:36] At the same time, the stock market was going so crazy over the AI wave that on 2nd of June 2026, Nvidia was worth $5 trillion and analysts were screaming to buy AI stocks.
[01:48] The AI will be a pretty [music] good thing to invest in.
[01:51] It is going to be, I think, just a booming year for AI and tech stocks, [music] especially in the first half of the year.
[01:57] But just 3 days later, something started cracking.
[02:00] Nvidia lost $320 billion in.
[02:04] Market cap.
[02:07] By 24th June, Micron was down by 13%, SanDisk was down by 10.59%, Apple fell by 6.1% and Soft Bank tanked 12%.
[02:14] On top of that, OpenAI delayed its IPO and slowly warnings are coming from the smartest people on earth are right now rising close to the same level in 2010.
[02:26] So what's the end?
[02:28] Is it a bubble that bursts eventually?
[02:29] I think it is.
[02:29] Yes.
[02:31] The problem with the AI capex boom is not only is it immense but a big chunk of it is funded with debt and that pain doesn't stay restricted.
[02:39] It spills over into the rest of society.
[02:42] The second thing that happens when people get very excited as they are today about artificial intelligence for example is every experiment gets funded.
[02:53] This is a kind of industrial bubble as opposed to financial bubbles.
[02:55] Now looking at this madness, I went back to understand all the bubbles in history.
[02:59] I read the Wall Street Journal, the Financial Times, the CNBC transcripts and even the JP Morgan.
[03:05] Gap analysis to understand why does Ray Dalio call this a textbook example of a bubble.
[03:10] And by the end of this video, you will understand better than 99% of Indian investors, whether this is the greatest business bet in human history or the greatest bubble ever inflated.
[03:18] Why are Michael Bur, Ray Dalio, and Jeff Bezos implying that this is a bubble?
[03:23] And what happens when this bubble bursts?
[03:25] Before we move on, if you're a manufacturer or if you run an e-commerce store, ODO has something very special for you.
[03:31] ODO understands that as a business owner, you are already juggling between three different tools.
[03:34] One for inventory, one for accounting, and one for sales.
[03:37] And every single day, something slips through the cracks even with your best efforts.
[03:40] Some order goes missing, the stock count doesn't match, the customer calls get lost between two teams, and hours are wasted just keeping these disconnected systems alive.
[03:48] Now, imagine if all of this ran from one place.
[03:52] This is exactly where ODO comes in.
[03:54] Odo is an all-in-one business management software that brings together 45 easy to use applications under one roof.
[04:01] So whether you are managing sales, invoicing, inventory, projects, or even building your own website, UDO runs.
[04:07] Everything from a single platform.
[04:09] So no more juggling and no more messy integrations.
[04:13] And the real magic is that every app talks to each other.
[04:15] So once you make a sale on your e-commerce store, the invoice is automatically created and your stock is updated in your inventory.
[04:21] So you always get a real-time view of your business in one place.
[04:25] And as your business grows, ODO grows with you.
[04:29] And the best part is you can get started with ODO for free today.
[04:31] And as your business scales, you can access the full suite for just 580 rupees per user per month.
[04:37] So if you want to grow your business in a flash, click the link in the description and start your journey with ODO today.
[04:44] [music]
[04:46] This is the story of one of the greatest bets in human history.
[04:48] Before we go anywhere, let me install a mental model in your head.
[04:52] Because if you don't understand what a data center actually is, none of the numbers will make sense.
[04:56] Imagine your phone.
[04:58] When you type a question into Chad Gupty, your phone doesn't just answer by [music] itself.
[05:02] Your phone is just a screen with Wi-Fi.
[05:04] The actual thinking happens somewhere else in a warehouse.
[05:06] A giant windowless.
[05:09] Industrial warehouse which is filled with metal racks.
[05:11] And each rack is installed with thousands of these [music] chips.
[05:15] That warehouse is called a data center.
[05:17] Each one of these buildings can hold 100,000 Nvidia GPUs.
[05:22] Each Nvidia GPU cost 30 to $40,000.
[05:26] So one building holds 3 to 4 billion worth of chips.
[05:29] One building just holds [music] 3 to 4 billion worth of chips.
[05:33] This 3 to 4 billion is [music] just for chips.
[05:40] On top of that, you have to power them, cool them, connect them with [music] high-speed cables and build them with concrete, security, and fire suppression.
[05:47] All in all, a single large AI data center cost 10 to 25 billion to build.
[05:53] [music] And this is where the race is happening.
[05:55] Like I told you in the data center case study, if you look at this graph, in 2010, the world created or replicated two zetabytes of data.
[06:03] That's roughly 2 trillion GB.
[06:06] But fast forward to today, something terrifying is happening.
[06:08] By 2026, the.
[06:11] [music] World is projected to generate 221 zetabytes of data.
[06:13] That's over 100x more than in 2010.
[06:16] So, the world is producing more data in a month than it did in all of history until 2010.
[06:22] This is the reason why the investment in data centers has shot up from $90 billion to $725 billion in just the last 6 years.
[06:33] Now, here's a number that made me question everything that is happening.
[06:37] The PIMCO report says that big tech capex will consume 94% of operating cash flows.
[06:42] I repeat 94% of operating cash flows over the next 2 years.
[06:46] You know what that means?
[06:49] If big tech earns $100, they will spend $94 back into building AI infrastructure.
[06:55] Only $6 will be left for dividends, buybacks, salary hikes, innovation, and everything else.
[07:04] In 2023, that same ratio was just 40%.
[07:07] And now it stands at 94%.
[07:10] So do you realize big tech is betting 94% of all?
[07:14] Its money into just one assumption.
[07:16] And the assumption says that in just 5 years, the world will need so much AI compute that every dollar being spent right now will practically look like a bargain.
[07:25] Sounds unstoppable, right?
[07:27] After all, we are producing so much data.
[07:29] Well, here's where it gets dangerous.
[07:31] Now, let's forget economics for a second and just imagine that you are a business owner.
[07:35] Let's say you spend $10 million building a coffee machine factory.
[07:38] Now, imagine that after all that, your factory only sells $400,000 worth of coffee machines in a year.
[07:45] So, you just make $400,000 from a factory that cost you $10 million.
[07:48] Is that good, bad, or terrible?
[07:51] You tell me.
[07:53] It's terrible, right?
[07:56] Why would you build another factory if your current factory doesn't make any money?
[07:57] Now take that exact same example and apply it to AI.
[08:00] Now let me show you the math.
[08:02] JP Morgan sat down and did this calculation and the logic is pretty simple.
[08:07] If you're an investor and you put money into something, you would at least expect a 10% return.
[08:12] That's bare minimum.
[08:14] Any serious investor demands on a risky bet like this.
[08:16] So JP Morgan said for AI giants to justify all the money that they're spending, how much money does AI actually need to bring in every year?
[08:25] The answer was $650 billion every single year.
[08:28] Okay, now remember this figure, $650 billion.
[08:31] Now, do you know how much AI is actually earning right now?
[08:34] Let's add it up.
[08:36] OpenAI, the makers of Chad GBT, make $25 billion a year, and they're losing $14 billion a year.
[08:43] Anthropic is set to make $47 billion at best if their current run rate goes on for one year.
[08:47] As of now, the target for Anthropic is about $26 billion by the end of this year.
[08:52] And let's say Gemini also makes $25 billion.
[08:54] So every major AI model company combined make around $75 billion with OpenAI losing 14 billion and Anthropic losing 3 billion in 2025 alone.
[09:06] Now put these three numbers side by side.
[09:08] Money that AI needs to earn to make sense $650 billion.
[09:12] Money AI is actually earning.
[09:16] $75 billion.
[09:19] Money that AI is losing is minimum $17 billion.
[09:21] But the money that the giants are spending on top of all of this is $725 billion.
[09:28] That difference between what they earn and what they need to earn is about 9 to 10 times.
[09:34] Now read that one more time slowly.
[09:36] For every single dollar that the AI industry is bringing in, the tech giants are spending 9 to 10 times more than they earn.
[09:44] This is why SEOA's David Khan calls this the $600 billion question.
[09:49] A $600 billion annual revenue deficit that nobody knows who will fill.
[09:53] Now the single biggest argument against this crazy number is Ganesh enterprises will pay money.
[09:57] Every single one of these companies will become profitable and investors will make money when the enterprises will pay money because AI is making all enterprises very very efficient at dirt cheap cost.
[10:07] Okay.
[10:11] Well, that is not the right argument because even I thought the same and then I found the service.
[10:15] McKenzie says 73%.
[10:18] Of enterprise AI deployments are failing to achieve projected return on investment.
[10:22] BCG says only 5% of companies are seeing substantial ROI from AI.
[10:26] MIT says there is a 95% failure rate in achieving measurable financial returns.
[10:31] Only 29% of the executives can even measure their AI return on investment.
[10:35] And this is where the story gets its first phase.
[10:38] Meet Flo.
[10:38] He runs an AI startup in San Francisco called Lindy.
[10:42] They have about 25 employees.
[10:45] In June 2026, he did an interview with CNBC that shook the AI industry.
[10:47] His team was spending more on Anthropic Cloud API than on their entire payroll.
[10:52] So, you know what Flo did?
[10:55] Flo switched 100% of his traffic to Deep Seek and his cost dropped by 90%.
[10:59] And then Uber CTO admitted publicly that Uber had blown its entire annual AI budget in just 4 months.
[11:06] And that ladies and gentlemen is the twist because everyone assumed that enterprises would keep paying more and more for AI tokens forever.
[11:13] That was the whole model.
[11:16] That is why OpenAI is worth.
[11:18] $850 billion.
[11:21] That is why Anthropic is worth $965 billion.
[11:24] But in June 2026, Enterprise started doing something that the market did not expect.
[11:28] They started looking for cheaper alternatives.
[11:30] Which is why Alex Karp, the CEO of Palanteer went on CNBC and said this on 1st of July.
[11:35] Every single enterprise I deal with, they're like, I am paying for tokens that create no value.
[11:42] These people are stealing the weights and alpha of my business and they're creating a wealth tax.
[11:47] And the reason for it is because [music] these models have been completely over irresponsibly oversold.
[11:51] And Palanteer, if you saw our previous case study, is one of the biggest software enterprise companies on earth.
[11:57] They sell to the CIA, the US government, Airbnb, JP Morgan, and god knows how many large companies.
[12:05] And the CEO of that company is telling you that something has completely gone wrong.
[12:09] Now, at this point, I know exactly what you're thinking.
[12:11] You must be thinking, "Yeah, Ganesh, this is a rich man's problem."
[12:14] Nvidia losing 500 billion, Sam, Dario, Sundar, they're all billionaires.
[12:18] How am I getting affected by all of?
[12:20] This?
[12:20] Well, let me take you to South Korea and show you how.
[12:26] This is a factory in South Korea that is owned by Samsung.
[12:27] This factory makes a very specific kind of memory chip called DM.
[12:32] The same DAM that goes into your laptop, your smartphone, your Xbox, and even your washing machine.
[12:37] In 2024, Samsung had a choice.
[12:39] It could sell its DAM to consumer companies like Apple, HP, or Dell.
[12:43] Or it could sell a special extremely expensive version called high bandwidth memory to AI data centers.
[12:50] And guess which one pays more?
[12:53] The AI data centers paid 10x more per module.
[12:56] So Samsung, SKH Highix and Micron, the three companies that control 90% of the world's memory chip supply, did what [music] any factory would do.
[13:06] They shifted 93% of their production towards AI memory because that is a rule of capitalism, right?
[13:13] Capital always flows to the highest bidder.
[13:15] Now watch what happens at bigger scales.
[13:17] DM prices are
[13:20] Up by 171% year-over-year as of March 2026.
[13:23] DDR5 memory are up 4x since September 2025.
[13:27] A contract price for PC memory is up by 105 to 110% in one quarter.
[13:32] In fact, Dell CEO said that the price of 1 GB of DAM went from 0.43 to $2.39 in just 6 months.
[13:37] That is a 5 and a half times increase in price.
[13:41] In fact, that is why on 25th of June 2026, Apple did something that it had never done before.
[13:47] In the middle of a product year, Apple simply raised their prices.
[13:50] This is the reason why they said, "We have never seen a competent price increase this much this quickly."
[13:56] We've shielded our customers from these increases so far.
[14:00] But now we've reached a point where we need to begin raising prices.
[14:02] Now that is Apple telling you this guys.
[14:04] The richest most vertically integrated tech company in the world is telling you that they cannot absorb this cost.
[14:10] That is how you are paying the AI tax.
[14:13] But this is where a scary question arises.
[14:16] If enterprises are moving off claw to save
[14:21] 90%.
[14:24] If Apple cannot absorb cost anymore, if the ROI is broken, then why are Amazon, Microsoft, Google, and Meta still spending more?
[14:32] Well, the answer is one of the most fascinating concepts in economics, and it explains every single bubble in human history.
[14:38] It's called the capital cycle.
[14:38] [music]
[14:40] In this cycle, there are four steps.
[14:42] Step number one, high returns attract capital.
[14:44] Step number two, capital keeps flowing until over capacity is built.
[14:49] Step three, return over capacity eventually results into collapse.
[14:54] And step four, everyone dies except a few survivors who eventually make a fortune when demand catches up.
[14:58] And every bubble in modern history has followed this exact same pattern.
[15:02] Let me show you how.
[15:05] In 1996, the US passed the Telecommunications Act because just like AI, the internet back then was a life-changing technology which was exploding in demand.
[15:13] The story was so intoxicating because it was clear to the world that internet was the future.
[15:17] Data traffic was exploding and everybody just knew that bandwidth.
[15:21] Demand would grow forever.
[15:24] Some founders even believed that internet traffic would double every 3 months.
[15:29] So money came pouring in to build the fiber optic cables.
[15:31] And then came the flood.
[15:33] Several companies raised to lay fiber optic cables across the country.
[15:37] And in just 5 years after that act, telecom companies poured more than $500 billion into cables, switches and networks.
[15:43] And if you look at the financials of these companies, you will see why the AI bubble is very similar.
[15:48] A company called Global Crossing went from a small equity check to a $47 billion valuation without ever making a single year of profit.
[15:56] Corvis, a fiber equipment startup, pulled off a $1.1 billion IPO with literally 0 in revenue and carried a $32 billion market cap.
[16:05] And just when everyone thought they'll become millionaires and billionaires, the collapse happened.
[16:08] You know what happened?
[16:10] Everyone thought that the internet will explode by 1,000% year on year, but the internet traffic only exploded by 100% year on year, which was great, but not great enough to justify the cost of investment.
[16:21] You know how.
[16:23] Much of this installed fiber was actually utilized?
[16:25] Take a guess.
[16:25] 50%, 20%, 10%, at least 5% must have been utilized, right?
[16:32] Well, guess what?
[16:34] By early 2000s, as little as just 2.7% of the installed fiber was actually carrying data.
[16:40] Over 95% sat unused underground.
[16:43] That is how trillions of dollars of cable got buried without earning anything.
[16:46] So when there was no revenue, bandwidth prices collapsed by up to 90% and the giant started failing.
[16:53] WorldCom, after hiding $3.8 billion of expenses to fake profits, filed the biggest bankruptcy in US history.
[16:57] Global Crossing, that $47 billion darling went bankrupt.
[17:01] And in total, the telecom crash wiped out $2 trillion of market value with stocks going down by 95%.
[17:08] And then came step four, the survivors.
[17:10] Now, here's where the twist comes in which makes it the perfect mirror for AI.
[17:14] Those fiber optic cables did not vanish.
[17:17] They stayed in the ground and within a few years, demand finally arrived.
[17:21] YouTube happened.
[17:23] Streaming started, cloud storage became a real thing, and [music] smartphone became popular.
[17:29] And suddenly the world needed exactly what had been overbuilt.
[17:32] So the survivors bought the wreckage for dirt cheap prices and that wasted cable became the physical backbone of the modern internet.
[17:38] The same infrastructure that made Google, Netflix [music] and AWS possible.
[17:42] So do you realize that technology was real?
[17:45] The internet did change everything but the bubble still burst.
[17:49] Why?
[17:49] Because the demand was exploding but not so much to justify over capacity.
[17:53] So the technology survived but the companies that built did not.
[17:57] Now, here's what the pattern looks like.
[17:58] Britain in 1846 authorized 9,500 miles of track and one/ird of it never got built.
[18:03] And then the bubble burst.
[18:05] America in 2000 laid millions of miles of fiber and 97% of it was unused and eventually the bubble burst.
[18:10] In 2026, America alone is building 725 billion of data centers per year and we don't know how much of it will actually be used.
[18:21] So the question is, will it all be worth it and become the greatest tech?
[18:24] Story ever told?
[18:27] Or will it go down as the greatest bubble in world history?
[18:29] Only time can give us the answer.
[18:32] So is this definitely a bubble?
[18:34] Well, we don't know that yet.
[18:35] Why?
[18:35] Because the companies in the telecom bubble were funded by debt and they were losing money.
[18:38] But Nvidia earned $120 billion in net income last year.
[18:41] And Microsoft, Google, and Amazon are literally the most profitable enterprises in human history.
[18:46] So they won't collapse like other weak companies.
[18:47] Similarly, at the 2000.com peak, the NASDAQ 100 forward PE was about 60x.
[18:53] Today, it's around 26x.
[18:56] It's higher than normal, but nowhere near the insanity of 1999.
[18:58] So, if anyone tells you for certain that this is a bubble, they're lying to you.
[19:03] And anyone tells you that it is definitely not a bubble is also lying to you because the truth is uncomfortable and it's somewhere in between.
[19:09] There is a very high possibility of a bubble, but not a certainty.
[19:13] The technology is real, the revenue is real, and we're not betting on whether AI changes the world or not.
[19:18] We are betting on whether the price for it actually makes sense or not.
[19:21] So now the question is what exactly is going to
[19:25] Happen if the bubble burst?
[19:26] And what if it doesn't?
[19:29] Well, there are two possibilities.
[19:31] Path one, the bubble pops, jobs are lost, the NASDAQ crashes, and every big tech company slams the brakes on spending, and that spending is what feeds our Indian IT and service sector.
[19:42] So your cousin's first coding job disappears before the boom can catch him.
[19:44] Path two is that the bubble doesn't pop.
[19:46] Instead, to justify those trillion dollar valuations, the company will try to race towards profit.
[19:51] So the price of AI, as in the token cost will shoot up and suddenly only the giants will be able to afford AI.
[19:57] So the cheap AI tools that you use today will eventually become a luxury.
[20:01] So a lot of AI products might die not because the tech failed, but because it just got too expensive to run.
[20:07] Or lastly, we could expect a miracle that will drop down the token cost, will make enterprises pay, and everyone will make money.
[20:14] But that, my dear friends, is a teeny tiny possibility.
[20:16] This, my dear friends, is the story of the AI bubble.
[20:19] Now, you tell me in the comments what do you think about the situation with the trillion dollar valuation that we seeing.
[20:23] Is this really a bubble or is
[20:25] this the greatest bet humanity has ever
[20:27] taken? That's all from my side for
[20:29] today, guys. If you learned something
[20:30] valuable from this case study, please
[20:32] hit the like button to support our work.
[20:34] And for more such business and political
[20:35] case studies, please subscribe to our
[20:37] channel. Thank you so much for watching.
[20:38] I will see you in the next one. Bye-bye.
[20:50] [music]