The Truth About the “AI Bubble”

Hi, Jeff’s managing editor here…

Before we hand things over to Jeff for today’s AMA, we just wanted to take a moment to thank everyone who joined Jeff and Larry on Wednesday to learn about what Larry is calling the last unrigged market in America…

As Larry discussed at the event – and as many of our readers already know – the biggest fortunes are often made by those who had the early-mover advantage.

Bitcoin’s early buyers had years of runway before the institutions arrived. The first options traders had an edge that lasted decades. Massive IPOs where the lion’s share of profit goes to those individuals and institutions that got in ahead of the public debut.

And the biggest firms see your trades before you place them. They’re faster than you. They skim a little off the top, millions of times a day. And the best opportunities remain locked behind closed doors.

It’s a game that is rigged against everyday investors.

That’s part of what makes Larry’s work and his strategy so exciting. He’s trading in a market where none of that is true. As he says, it’s “the last unrigged market in America.”

And using his strategy, everyday investors can use this market to pull regular gains in the double and triple digits.

It was a great time, and we hope everyone who attended enjoyed the briefing.

And for those who couldn’t make it – or who may be interested in hearing more about Larry’s strategy – we do still have a replay available.

You can go here to access it, then read on for today’s AMA from Jeff.

The AI “Bubble”

Dear Jeff,

I would highly appreciate your ASAP response, because the urgent issue summarized below is certainly important not only for me, but also for most of your other subscribers…

A few months ago, some experts explained in detail why the circular financing scheme around OpenAI most probably would lead to the burst of the AI bubble.

Later on, several experts also explained that first a ‘selective’ melt-up will come in the tech sector, but probably within 6 to 12 months a devastating market crash is highly probable.

During the last weeks, you (and many other investment experts) suggested investing in several microcap and midcap companies before the SpaceX IPO, because their stocks were expected to surge on the tailwind of SpaceX. However, during the last couple of days, many of these stocks were beaten.

Today, Jim Rickards announced that, in his view, the AI bubble may burst any day.

I (and most probably your subscribers, too) would highly appreciate your opinion on this issue, also including the suggested preparations.

Thanks for your time and help in advance.

Best regards.

– Attila K.

Hello Attila,

The premise that we are in an “AI bubble” right now is fundamentally wrong. I feel embarrassed for those “experts” who proclaim there is a bubble.

I was years ahead of this trend, and consistent with my predictions that Wall Street was way underestimating the amount of capital expenditure spend that would take place to build out the AI data centers required to achieve AGI, run AI applications, and ultimately make it to ASI.

Every quarter, the forecasts from the major investment banks continue to increase. They are still trying to wrap their heads around what is happening.

The business of artificial intelligence is real… It’s grounded, and it’s the definition of exponential growth. OpenAI is forecasting about $30 billion in revenue this year. That’s up from a mere $1 billion in revenue in 2023.

Anthropic looks like it will hit $10.9 billion in revenue this quarter, meaning that it will exceed OpenAI revenues for the full year.

I could go down the list, one after the other. The business growth is extraordinary, and there is a fundamental need to grow the AI infrastructure as quickly as possible to keep up.

It is critical to note that the majority of capital expenditures are being driven by massive companies that have the money to spend (as opposed to speculative leverage). Microsoft (MSFT), Amazon (AMZN), Alphabet (GOOGL), Meta (META), OpenAI, NVIDIA (NVDA), and SpaceX (SPCX) have the capital to make these investments.

Yes, companies like NVIDIA or AMD (AMD) are investing in some AI companies as part of a deal to sell more of their semiconductors. But they have the capital to do so as they are gushing free cash flow.

We should keep things in perspective right now. It has only been five trading days since the extremely successful SpaceX IPO. SpaceX is roughly tied now with Amazon as the fifth most valuable company in the world.

The debate over SpaceX right now is crazy. It reminds me of Tesla. Most analysts valued Tesla as a car company. I always valued it as an AI company. And despite everything that Tesla has accomplished to date, there are still some who value it as a car company.

The same is true of SpaceX. Many are valuing the company as a launch services company. And yet, Starlink is the largest percentage of its current business, and AI will soon be the largest overall part of its business. SpaceX became an AI company after it acquired xAI and developed its strategy for AI data center satellites. SpaceX and Musk have been open and clear about the strategy, and yet many still think SpaceX is just a rocket company.

My point is that many still don’t see the obvious today. The investment thesis will play out. It has only been five trading days. What will happen in the weeks and months that follow will be extraordinary.

I recommended Micron Technology (MU) about 18 months ago in The Near Future Report when I explained that the market didn’t understand how critical high bandwidth memory would become for AI data centers and applications. My subscribers are now up 915% on that recommendation.

There will eventually be a bubble and a melt-up, but that time is not now. When such a time comes, my subscribers will be the first to hear about it.

Recommended Links

New AI-Powered Search Engine Finds Stocks With 10X Gain Potential

This new AI-powered platform matches today's unknown stocks to yesterday's biggest winners. In back-testing, it would have helped you find gains 17X higher than Nvidia... 5X higher than Alphabet... and 18X higher than Amazon – over the exact same time frame. You can unlock temporary access to a special "beta" version of this new AI breakthrough, free of charge.

See this Building the Size of Several Football Fields?

This is where Elon Musk is housing an AI technology that Jeff Brown believes will help power the next monster IPO on Wall Street.  You see, while everyone was distracted by the SpaceX IPO… Elon Musk quietly started backing a NEW AI startup that has been called…  “The fastest-growing business in the history of capitalism.”  Click here to get the name completely free of charge…  And Jeff will also show you how to claim a stake for as little as $50.

Elon Musk’s X Money Launch

What do you know about the bank that Elon is planning to launch shortly after the SpaceX IPO is done? From what I have heard, which isn’t much, it will be this year also, and since it will be an all-in-one financial institution, there is one company that will be in the center of it all. Looking forward to hearing what info you have on this.

– Elizabeth T.

Hi Elizabeth,

There is no indication at all that Musk will launch a bank. What he is in the process of launching is X Money, which is a full-featured payments platform embedded into the X application. X uses Cross River Bank (private) as its banking partner for the back end of its financial services.

X Money is already in the process of being rolled out gradually. The goal is, of course, to take it nationwide in the U.S. and then expand from there.

X is another one of Musk’s businesses that is being heavily discounted by the market. Its social media platform is worth a fortune, and X Money will give the platform that much more utility, which will dramatically increase its value.

I’m quite excited to see how X Money develops later this year. It’s no easy task to get the approvals to transmit money in all 50 states in the U.S. But once those financial rails are in place, what X can do with that is something incredible.

Chinese Independence from Western AI Infrastructure?

While earlier DeepSeek models relied heavily on Nvidia hardware – and as such, indirectly on Nvidia’s CUDA ecosystem – [its newest V4 model] is the startup’s first to operate with significantly greater independence from Western technology. This is great for Chinese tech giants, as V4 is also based on open source. Tencent, ByteDance, and other [Chinese tech] companies can now copy the code and customize it based on their specific needs. As a result, Chinese hyperscalers are practically no longer reliant on Western chips and infrastructure.

How might this Chinese breakthrough impact the US market for chips, etc.? I am a long-term customer and appreciate your valuable services, particularly regarding IT and AI technologies. Best regards.

– Wolfgang B.

Hi Wolfgang,

It’s important to put this topic into context so that we’re all on the same page.

DeepSeek models relied entirely on NVIDIA hardware, just not in the way that most think. DeepSeek “stole” critical data and outputs from U.S. AI frontier models that were entirely trained on NVIDIA hardware.

DeepSeek gained API access to OpenAI’s frontier AI models and used obfuscated accounts to avoid restrictions. OpenAI publicly shared that DeepSeek employees deployed programmatic code that was used to harvest the outputs of those models at scale to develop its own model. These facts were corroborated by Microsoft.

Anthropic was a victim of DeepSeek in a very similar way. DeepSeek developed prompts to generate outputs and high-quality synthetic training data for its own model. Anthropic discovered about 24,000 fraudulent accounts that were used to obfuscate this theft. Anthropic referred to this as “industrial-scale distillation attacks.”

This is intellectual property theft, plain and simple. And the technology that it stole was entirely built on NVIDIA hardware.

Now, where there is some truth to your comments is that Huawei – another China-based company that was founded on intellectual property theft from U.S. companies – has developed its Ascend semiconductors, which are designed for inference – the running of AI applications.

DeepSeek can now use Ascend semiconductors to run its AI model, but DeepSeek is still heavily reliant on NVIDIA or AMD hardware for training new frontier models.

These points also apply to any companies that copy DeepSeek V4 and run their own models. They can use the Ascend chips, but they still need NVIDIA or AMD for training new models.

Chinese companies do have obfuscated ways to gain access to NVIDIA hardware through proxies, but it is nearly impossible to do at the scale necessary to compete with anything that is happening in the U.S.

Another aspect to keep in mind is that all manufacturing capacity for NVIDIA, AMD, and others is spoken for at least a year out. And the capital expenditure forecasts for U.S. AI data center construction will exceed $1 trillion next year and be even higher the year after that.

Put simply, the U.S. market is dominating in frontier AI model development, AGI, agentic AI, and AI infrastructure build-out. That will not change anytime soon.

And by late 2028, large-scale AI data center satellites will begin orbiting the Earth, leading to the next wave of AI infrastructure build-out, ensuring that the lack of available energy on Earth doesn’t slow down technological advancement for U.S.-based companies.

The world remains heavily dependent upon U.S. semiconductors for AI and U.S. AI infrastructure. And the U.S. has the strongest financial markets in the world to ensure that any high-quality project or company is getting the funding needed to advance artificial intelligence.

That’s all for this week’s AMA. As always, you can reach me or my team right here.

Have a great weekend, everyone.

Jeff