WHEN INVESTING BECOMES A LIFESTYLE YOU WEAR IT!

NEW MERCH ALERT - WEALTHY RED - Limited Quantities - Grab Yours Today! CLICK HERE

I am taking taking a poll:

How much would you pay for a research that not only stopped you from making the wrong decision but taught you how to make the right decision? I am trying to establish a price for the value. The way it works is you send in a ticker with your payment and you will receive the most comprehensible actionable research report you have every seen, even equipped with price predictions based off free cash flow. Just reply with the amount.

Thank you in advance. KD

Wall Streets Rage Against The AI Machine…

The market keeps acting like its full-time job is to manufacture reasons to fade Nvidia, then looks shocked when the earnings show up like a debt collector.  First it was “Google will eat their lunch,” then “Copilot disappointment means the AI capex party is over,” as if the Street is desperately trying to talk the stock back into 2022 prices so it doesn’t have to admit it whiffed on the entire AI regime change.

Do you really believe AI is a bubble or do you want us to believe its a bubble while you make out like “The Short Side Bandits”.

The pattern is almost comical at this point:

First “Google TPUs will make Nvidia obsolete.” Reality: Google is ramping TPUs, sure, but it still buys Nvidia GPUs and no one has a credible forecast where Nvidia loses its leadership this decade.

And then “A Copilot miss means hyperscalers will slash GPU spending.” The reality is Microsoft, Amazon, Google, and Meta are all guiding AI and data center capex into the hundreds of billions over the next few years; they talk “discipline” on TV and then quietly pour concrete and install more racks.

Meanwhile, Nvidia just printed a quarter with roughly $57 billion in revenue, $51 billion of that from data centers alone, up more than 60% year over year, and guided higher again.  That is not story-stock behavior; that is “this is the toll road” behavior.

This is what I call Dot-com PTSD and here is why the Street is gun-shy, a lot of today’s senior money managers and strategists either lived through, or were trained by people who lived through, the dot-com bubble—and they got cooked.  Back then, internet stocks went vertical, the NASDAQ’s P/E blasted past 90, IPOs from profitless companies flooded the market, and something like 70–80% of new listings had negative earnings.

And Wall Street was absolutely complicit. Underwriting desks shoveled hundreds of internet IPOs out the door from 1996–2000, analysts slapped Buy ratings and sky-high price targets on businesses with no cash flow, and everyone collected fees until the music stopped and trillions in paper wealth evaporated.  So now, AI shows up with real revenue, real margins, and real infrastructure spend—and instead of learning the right lesson (“don’t overpay for zero earnings”), they generalize the wrong one (“all tech booms are scams”).

This time, the earnings are already here and the big difference is Nvidia’s earnings are not a promise; they are already on the income statement.

• Data center revenue has exploded from around $15 billion in fiscal 2023 to an estimated near-$200 billion run rate by fiscal 2026.

• In the latest quarter, data center was nearly 90% of total revenue, grew over 60% year over year, and Nvidia guided to another massive sequential jump.

At the same time, media and some “AI bubble” commentators happily pump up story stocks and concept names with thin or no earnings—exactly the dot-com playbook—while nitpicking the one company that is actually minting cash selling the shovels. 

Deals with OpenAI and others are being talked about in terms of hypothetical future software upside, but the current, hard-cash GPU orders that enable those futures are hitting Nvidia’s P&L right now.

 Wall Street’s is framing a no-win set-up, even Jensen Huang had to tell employees they’re in a “no-win” setup: if Nvidia ever misses, it’s “proof” of an AI bubble; if Nvidia crushes numbers, it’s “fueling” an AI bubble. 

That circular logic lets the Street be “right” no matter what happens, which is convenient if you missed the move and need to save face on TV.

The result is a bizarre split-screen: on one side, hyperscalers and sovereigns committing hundreds of billions to accelerated compute, data centers, and energy; on the other, pundits acting like this is Pets.com 2.0 while ignoring the fact that Nvidia is already printing tens of billions per quarter in net income from this buildout. 

AI isn’t the boy who cried wolf; it’s the guy at the door with a ledger and a payment schedule saying “adapt or die,” and the cash flows are already testifying.

Analysts living in some sought of selective reality distortion field. If this were a courtroom, the evidence against the Street would be straightforward:

• Exhibit A: During the dot-com era, analysts and banks promoted and underwrote wave after wave of unprofitable IPOs, helping drive the NASDAQ to absurd valuations before it collapsed.

• Exhibit B: Today, many of those same institutions are quick to scream “bubble” at the one AI company delivering gigantic, repeatable earnings while still happily backing speculative, low-earnings names and riding the multiple expansion there.

So when the market keeps spinning new stories—“Google is coming,” “Copilot spend is dead,” “AI capex is about to fall off a cliff”—it looks less like sober analysis and more like a coping mechanism from people who missed the boat and now need Nvidia to be wrong so they can feel right.  The problem for them is that cash doesn’t care about feelings, and right now, the cash is lining up on Nvidia’s side of the ledger.

 WHEN INVESTING BECOMES A LIFESTYLE YOU WEAR IT!

NEW MERCH ALERT - BILL GATES BLACK - Limited Quantities - Grab Yours Today! CLICK HERE

Pilot To Copilot We are Going Down In Flames…

Microsoft’s Copilot “miss” is the headline, but the real story is that the cloud money machine is on fire, capacity is tapped, and hyperscalers are in an arms race that makes cutting AI capex about as realistic as a diet during Thanksgiving.  That dynamic turns NVIDIA GPUs into the pickaxes in a gold rush, while the market keeps pretending one quarter of Copilot attach rates is the master valuation metric.

The narrative is “Copilot isn’t scaling fast enough, AI monetization is disappointing, time to question the AI spend.”  If the analysts left the ivory tower for one ,moment, they would see that Microsoft’s cloud and AI-related capex has exploded into the tens of billions per quarter, while Azure keeps compounding at high-20s to 40% growth with AI adding multiple points of incremental growth.

Investors are screaming about a software SKU while ignoring that cloud is more than half of Microsoft’s revenue and is still gaining share in a global market already pushing toward a trillion dollars.  Even if Copilot underwhelms near term, the workloads that matter—LLM hosting, inference, fine-tuning, and every “AI inside” enterprise app—still have to live on Azure, and that’s where the actual cash compounding happens.

The cloud is sold out, not overbuilt, and Microsoft is literally telling you demand is running ahead of capacity. Azure growth includes a fat contribution from AI services, and management has said AI demand is higher than available capacity, forcing them to pour about $19 billion into capex in a single recent quarter just to try to catch up.  They’ve even had to lean on third-party partners to extend Azure’s AI capacity because internal buildouts can’t keep up.

This is not what excess-bubble infrastructure looks like; this is a capacity-constrained toll road where the traffic jam is revenue waiting to be recognized.  Shortages of AI-optimized capacity are exactly why Nvidia’s higher-end GPUs have enjoyed long lead times and premium pricing, even as supply has started to improve from “forget it” to merely “wait a couple months.”

When hyperscalers say cloud and AI capex is “nearly all” of spend, what they mean in plain English is: “We’re buying data centers and GPUs like there’s no tomorrow.”  Demand for NVIDIA’s H100/H200 class GPUs has been so intense that lead times were measured in many months and still often stretch multiple weeks, even after supply ramped and secondary markets emerged.

On top of that, you now have sovereign funds, oil states, Fortune 500s, startups, and AI labs all hoarding these chips as strategic assets, pushing demand well beyond hyperscaler budgets alone.  So while commentators obsess over Copilot’s ASP and seat penetration, the actual bottleneck is still: “Can you get enough high-end GPUs and power to stand up AI clusters at scale, and can you do it before your competitors?”

Here’s where the market gets it backwards: pundits yell “AI bubble” and then in the next sentence admit hyperscaler and big-tech AI capex is projected to soar to well over $600 billion in 2026, after already doing over $200 billion in 2024.  Microsoft alone has ramped quarterly capex into the mid-30 billions recently, guided to even higher levels as it races to add AI capacity globally.

The idea that a wobbly quarter of Copilot metrics would force them to slam the brakes on this buildout ignores the actual game being played: this is a multi-year infrastructure land grab where whoever owns the most performant AI cloud wins the next decade of software.  You do not voluntarily stop pouring concrete and installing GPUs while Amazon, Google, and Meta are still laying track.

Analysts estimate total big-tech AI and infrastructure capex (largely these four) is running toward roughly $600 billion a year and climbing fast, with projections pushing to the trillions by 2030 in the next couple of years.  Microsoft is spending tens of billions per year on AI-oriented data centers and custom and third-party accelerators; Amazon is doing the same with a mix of Nvidia GPUs and its own Trainium/Inferentia chips; Google is leaning hard on its TPU roadmap while still buying Nvidia; and Meta is dropping tens of billions to stand up data centers for its LLAMA and recommendation workloads.

If any one of them truly “pulled back” on GPU and data center capex because one branded AI product wasn’t growing fast enough, that would not be “disciplined capital allocation,” it would be surrender.  The reality is they are locked in a prisoners’-dilemma-style race where each is forced to keep spending or risk losing share in AI workloads that will sit on these platforms for years, long after the market has moved on from trading headlines about Copilot attach rates.

The market is treating Copilot like a SaaS stock and ignoring that it is really a marketing wrapper for an underlying AI infrastructure build that is still starved for capacity.  The mispricing comes from focusing on near-term AI “features” instead of the long-duration value of owning the compute layer and the very scarce GPUs that actually power all of it.

That disconnect is exactly why Nvidia’s GPUs are still a hot commodity and why hyperscalers’ “we might slow down” body language rarely matches their actual capex line.  The bubble isn’t that they are spending too much; it’s that investors think they have the option to stop. They don’t—because in this game, losing the capex war means losing the cloud, and losing the cloud means losing the future cash flows everyone’s pretending to model off one quarter of Copilot.

WHEN INVESTING BECOMES A LIFESTYLE YOU WEAR IT!

NEW MERCH ALERT - BUY THE DIP NAVY - Limited Quantities - Grab Yours Today! CLICK HERE

Quick Links…

Finish Strong Brother…

What’s In Your 401k…

So Netflix Wants To Own The Bunny…

Thank you for reading, we appreciate your feedback—sharing is caring.

Kevin Davis Founder of Investment Dojo and Author of The C.R.E.A.M. Report