Why Cheap AI Just Triggered a Massive Tech Boom
- Rendement de la vidéo
- +2,72%
- Appels
- 11
- Achat / Vente
- 11 0
- Publié
Recommandations
L'entrée est le cours de clôture de l'actif à la date de publication. Le cours actuel est la dernière clôture enregistrée.
-
Entrée $357,16 03 sept 2026Actuel $364,38 09 sept 2026Résultat +$7,22vs. indice +3,4% SPY −1,4% sur la même période
Citer cette recommandation Voir la vidéo source * †
Contexte de la transcription source
…matter whether the winning chip is designed by Nvidia, AMD or customuilt inside Google or Meta. Every advanced piece of silicon on Earth must be etched in TSMC's foundaries and packaged with their proprietary advanced packaging technology. Third is Broadcom. As open models make intelligence cheap, massive tech giants are designing custom application specific chips to lower their operational bills. Broadcom is the premier partner that co-designs those custom chips while simultaneously dominating the high-speed Ethernet switching silicon that connects enterprise data centers. Fourth is advanced micro devices. AMD is not just an altern…
Third is Broadcom. As open models make intelligence cheap, massive tech giants are designing custom application specific chips to lower their operational bills.
-
Entrée $456,16 03 sept 2026Actuel $521,10 09 sept 2026Résultat +$64,94vs. indice +15,6% SPY −1,4% sur la même période
Citer cette recommandation Voir la vidéo source * †
Contexte de la transcription source
…pplication specific chips to lower their operational bills. Broadcom is the premier partner that co-designs those custom chips while simultaneously dominating the high-speed Ethernet switching silicon that connects enterprise data centers. Fourth is advanced micro devices. AMD is not just an alternative compute provider. Their server processors handle the massive system orchestration required to manage autonomous agents. AMD raised its long-term server market growth forecast from 18% up to 35% a year, adding 60 billion to their projections precisely while …
Fourth is advanced micro devices. AMD is not just an alternative compute provider.
-
Entrée $958,16 03 sept 2026Actuel $1 027,77 09 sept 2026Résultat +$69,61vs. indice +8,7% SPY −1,4% sur la même période
Citer cette recommandation Voir la vidéo source * †
Contexte de la transcription source
…dle the massive system orchestration required to manage autonomous agents. AMD raised its long-term server market growth forecast from 18% up to 35% a year, adding 60 billion to their projections precisely while open models were expanding. Fifth is Micron Technology. Generative intelligence cannot operate without massive memory bandwidth. High bandwidth memory is mechanically brutal to manufacture and the entire global production capacity across the industry is effectively locked up and sold out for quarters in advance. Sixth is Veritive Holdings. You cannot run 40 trillion…
Fifth is Micron Technology. Generative intelligence cannot operate without massive memory bandwidth.
-
Entrée $268,83 03 sept 2026Actuel $262,89 09 sept 2026Résultat −$5,94vs. indice −0,8% SPY −1,4% sur la même période
Citer cette recommandation Voir la vidéo source * †
Contexte de la transcription source
…gence cannot operate without massive memory bandwidth. High bandwidth memory is mechanically brutal to manufacture and the entire global production capacity across the industry is effectively locked up and sold out for quarters in advance. Sixth is Veritive Holdings. You cannot run 40 trillion tokens a day through highdensity servers without generating staggering amounts of heat. Traditional air conditioning cannot handle modern server racks that draw upwards of 100 kilowatts, making Vertiv's liquid cooling and power distribution systems an absolute physical requirement. Seventh is Palunteer technologies. Cheap mod…
Sixth is Veritive Holdings. You cannot run 40 trillion tokens a day through highdensity servers without generating staggering amounts of heat.
-
Entrée $182,53 03 sept 2026Actuel $169,53 09 sept 2026Résultat −$13,00vs. indice −5,7% SPY −1,4% sur la même période
Citer cette recommandation Voir la vidéo source * †
Contexte de la transcription source
…thout generating staggering amounts of heat. Traditional air conditioning cannot handle modern server racks that draw upwards of 100 kilowatts, making Vertiv's liquid cooling and power distribution systems an absolute physical requirement. Seventh is Palunteer technologies. Cheap models are completely useless to an enterprise unless they can be integrated securely into messy corporate data without making things up. Palanteer's ontology provides the enterprise operating system that turns cheap commoditized intelligence into real measurable corporate cash flow. Let's go over the bull and bear cases. To hold these companies with absolute conviction, you…
Seventh is Palunteer technologies. Cheap models are completely useless to an enterprise unless they can be integrated securely into messy corporate data without making things up.
-
Entrée $285,05 03 sept 2026Actuel $293,90 09 sept 2026Résultat +$8,85vs. indice +4,5% SPY −1,4% sur la même période
Citer cette recommandation Voir la vidéo source * †
Contexte de la transcription source
…ast 3 to four years and utility interconnection cues are backed up across the entire country. This means the next explosive wave of capital is flowing straight to the companies that generate base load electrons and build the physical grid. On the generation side, look at independent power producers like Constellation Energy, ticker CEG, and Vistra, ticker VST. These companies own massive fleets of nuclear and gas plants capable of providing dedicated round-the-clock power directly to hypers scale data centers without touching the public grid. On the physical grid equipment side, look at GE Verno…
On the generation side, look at independent power producers like Constellation Energy, ticker CEG, and Vistra, ticker VST.
-
Entrée $144,22 03 sept 2026Actuel $151,10 09 sept 2026Résultat +$6,88vs. indice +6,2% SPY −1,4% sur la même période
Citer cette recommandation Voir la vidéo source * †
Contexte de la transcription source
…ast 3 to four years and utility interconnection cues are backed up across the entire country. This means the next explosive wave of capital is flowing straight to the companies that generate base load electrons and build the physical grid. On the generation side, look at independent power producers like Constellation Energy, ticker CEG, and Vistra, ticker VST. These companies own massive fleets of nuclear and gas plants capable of providing dedicated round-the-clock power directly to hypers scale data centers without touching the public grid. On the physical grid equipment side, look at GE Verno…
On the generation side, look at independent power producers like Constellation Energy, ticker CEG, and Vistra, ticker VST.
-
Entrée $941,84 03 sept 2026Actuel $951,04 09 sept 2026Résultat +$9,20vs. indice +2,4% SPY −1,4% sur la même période
Citer cette recommandation Voir la vidéo source * †
Contexte de la transcription source
…tellation Energy, ticker CEG, and Vistra, ticker VST. These companies own massive fleets of nuclear and gas plants capable of providing dedicated round-the-clock power directly to hypers scale data centers without touching the public grid. On the physical grid equipment side, look at GE Vernova, ticker GEV, and Eaton Corporation, ticker ETN. GE Vernova builds the heavy duty gas turbines and grid electrification gear that utilities desperately need to expand capacity. Eaton practically owns the industrial switch gear, circuit breakers, and power distribution hardware required i…
On the physical grid equipment side, look at GE Vernova, ticker GEV, and Eaton Corporation, ticker ETN.
-
Entrée $397,12 03 sept 2026Actuel $415,22 09 sept 2026Résultat +$18,10vs. indice +6,0% SPY −1,4% sur la même période
Citer cette recommandation Voir la vidéo source * †
Contexte de la transcription source
…tellation Energy, ticker CEG, and Vistra, ticker VST. These companies own massive fleets of nuclear and gas plants capable of providing dedicated round-the-clock power directly to hypers scale data centers without touching the public grid. On the physical grid equipment side, look at GE Vernova, ticker GEV, and Eaton Corporation, ticker ETN. GE Vernova builds the heavy duty gas turbines and grid electrification gear that utilities desperately need to expand capacity. Eaton practically owns the industrial switch gear, circuit breakers, and power distribution hardware required i…
On the physical grid equipment side, look at GE Vernova, ticker GEV, and Eaton Corporation, ticker ETN.
-
Entrée $228,45 03 sept 2026Actuel $223,67 09 sept 2026Résultat −$4,78vs. indice −0,7% SPY −1,4% sur la même période
Citer cette recommandation Voir la vidéo source * †
Contexte de la transcription source
…se they absolutely will. The real question is which companies sit at the physical toll booth extracting cash regardless of which software model wins the race. Seven businesses hold a virtual chokeold on this entire infrastructure pipeline. First is Nvidia. The bears obsess over raw chip design, but they completely ignore the proprietary NVLink interconnect and Spectrum X networking that tie thousands of chips into a single giant brain. When millions of agents run simultaneously, moving data between chips is the ultimate bottleneck and Nvidia owns that networking standard. Second is Taiwan semiconductor. It does not matter whether the winning chip is designed by Nvidia, A…
First is Nvidia. The bears obsess over raw chip design, but they completely ignore the proprietary NVLink interconnect and Spectrum X networking that tie thousands of chips into a single giant brain.
-
Entrée $417,01 03 sept 2026Actuel $435,36 09 sept 2026Résultat +$18,35vs. indice +5,8% SPY −1,4% sur la même période
Citer cette recommandation Voir la vidéo source * †
Contexte de la transcription source
…NVLink interconnect and Spectrum X networking that tie thousands of chips into a single giant brain. When millions of agents run simultaneously, moving data between chips is the ultimate bottleneck and Nvidia owns that networking standard. Second is Taiwan semiconductor. It does not matter whether the winning chip is designed by Nvidia, AMD or customuilt inside Google or Meta. Every advanced piece of silicon on Earth must be etched in TSMC's foundaries and packaged with their proprietary advanced packaging technology. Third is Broadcom. As open models make intelligence cheap, massive tech giants are designing custom application specific chips to lower their operational bills. Broadcom is the premier partner that co-designs those custom chips while simul…
Second is Taiwan semiconductor. It does not matter whether the winning chip is designed by Nvidia, AMD or customuilt inside Google or Meta. Every advanced piece of silicon on Earth must be etched in TSMC's foundaries and packaged with their proprietary advanced packaging technology.
Transcription Complète
Last year, the financial media told you the party was officially over for American tech. A free Chinese model hit the web and in less than 24 hours, 589 billion was erased from Nvidia in the single biggest market cap wipeout in history. The consensus said expensive chips were dead. The big tech moat was gone and everyone holding semiconductor stocks was left holding an empty bag. We've all sat in front of our screens watching a position bleed 15% in a single morning. Feeling that immediate pit in our stomach. You start wondering if you missed something obvious or if the entire thesis you built your portfolio around just collapsed under a single headline. It makes you want to hit the sell button just to make the pain stop, which is exactly what the herd did. What Wall Street completely missed that morning was worked out back in 1865 by an economist staring at piles of coal. The very thing everyone thought would kill the industry is the exact fuel that makes it 10 times larger. Today we are breaking down why cheaper artificial intelligence is an absolute gold mine for physical tech and the seven fortress companies built to harvest every dime of it. And if you stick around past the wrap-up today, I left a little reward waiting for you at the very end, strictly for the real ones who stay until the screen goes black. I'm an investor, not a financial adviser, so do your own research. But let's cut straight through the noise and look at the real mechanics. The knee-jerk reaction on trading desks assumes that if software gets cheaper to make, total spending on hardware must collapse. That sounds logical on the surface, but it completely falls apart when you look at the real history of computing. When roomsiz mainframes cost millions of dollars, only governments and military labs could afford to run them. The moment microprocessors made computing dirt cheap, the market did not shrink down to zero. It exploded into billions of personal computers, smartphones, smartwatches, and connected cars. The exact same illusion is playing out right now across every financial news channel. They see a software model that costs less to run and assume data center spending is finished. In reality, collapsing unit costs do not destroy demand. They blow the doors wide open. Back in 1865, a British economist named William Stanley Javons published a study on the United Kingdom's coal supply. James Watt had just built an engine that burned coal far more efficiently than anything that came before it. Naturally, the smartest minds in London predicted Britain would burn dramatically less fuel because every engine needed less coal to do the job. The exact opposite happened. Because steam power suddenly became cheap, factories that could never afford an engine began buying them by the dozens. Entire industries mechanized overnight and total coal consumption surged to all-time highs. That economic pattern is known as Javon's paradox. When a basic resource becomes dramatically more efficient and cheaper to use, total consumption does not drop, it skyrockets. We have all fallen into the trap of thinking a drop in production cost means a drop in business revenue. You see a headline about open- source code catching up to closed labs and you feel that sudden urge to dump your long-term positions. We tell ourselves the moat is gone. We forget the underlying math and we mistake a price cut for a collapse in total volume. That's the trap retail investors step into every single cycle. When an artificial intelligence query costs a fraction of a cent, you do not just run fewer queries to save a dollar. You take that intelligence and shove it into every workflow, every piece of software, and every device on the planet. Take a trip back to January 27th of last year when Deepseek dropped their R1 model out of China. It matched Elite American Reasoning benchmarks at a reported fraction of the original training cost. The algorithmic training desk saw that headline, hit the panic button, and triggered a violent sell-off across the entire chip complex. Nvidia stock dropped 17% in a single trading session, wiping out more than half a trillion dollars of value in six hours. The broad semiconductor indexes logged their worst single day route since the market shut down in March of 2020. The prevailing media narrative was that spending tens of billions on advanced hardware was a total waste of capital. The engineers who actually run these clusters immediately pushed back. They pointed out that training a model is only the entry ticket to the game. Running that model for millions of live users around the clock requires an absolute mountain of computing power and ultra high-speed networking. Wall Street traded the headline while the people building the systems looked at the unit economics. The market panicked because it thought the race was about who could build the cheapest model. They completely missed the fact that the cheaper the model gets, the more hardware the real economy consumes to run it. Let's strip away all the technical jargon and make this so dead simple a caveman could understand it. Think of a token like a single word and think of computer chips like an engine that burns gasoline. Every time an artificial intelligence reads or writes a word, it burns a drop of fuel. In the old days, you opened a chat window and asked for a chocolate chip cookie recipe. The machine spat out 200 words, burned a teaspoon of gas, and went right back to sleep. That is basic conversational computing and it barely moves the needle on data center power. Now look at what's happening today with autonomous agents. You do not ask an agent for a recipe. You tell it to audit 50,000 corporate tax returns, write the code, fix the errors, and file the paperwork. That is not a quick chat. That is a digital worker laboring 24 hours a day without taking a lunch break. Instead of burning a teaspoon of gas, that single digital worker burns through an entire tanker truck of fuel every afternoon. Look at the independent inference providers serving these cheap models right now. One single provider is already processing over 40 trillion tokens every single day. That is double the volume of the biggest proprietary platform on the planet and it is growing exponentially. Demand did not vanish when models became cheap and open. It moved to cheaper pipes and exploded into a raging fire hose of volume. Every single word that gets processed has to run across physical silicon backed by physical memory cooled by physical water. That explosion in token volume is exactly why long-term investors need a framework, not just stock picks. Inside my Patreon, I share every trade I take and the reasoning behind it, so you can see how a thesis like this actually gets executed. The link is below if you want to see how we operate. But let's look at why you cannot just buy these blindly because the volatility will absolutely shake you out. None of this technology buildout happens in a neat orderly straight line. The market is a manic depressive machine that overreacts to every single press release, quarterly earnings rumor, and geopolitical headline. If you want the life-changing upside that comes with owning generational monopolies, massive volatility is simply the price of admission. I have a lot of money invested across this hardware ecosystem, which means I cannot afford to trade on wishful thinking or panic on a red day. When you look back at that catastrophic January session where Nvidia shed 17%, look at where the stock is trading today. The people who panic sold at the bottom handed their shares directly to patient institutional capital at a massive discount. When you own the structural bottlenecks of a multi-t trillion dollar shift, macro pullbacks are not a signal to run away. They are the rare windows where the market offers you prime assets on sale because the crowd cannot see past next week. You do not build lasting wealth by dumping your portfolio every time a competitor ships an open-source model. The question is not whether open models will get faster and cheaper, because they absolutely will. The real question is which companies sit at the physical toll booth extracting cash regardless of which software model wins the race. Seven businesses hold a virtual chokeold on this entire infrastructure pipeline. First is Nvidia. The bears obsess over raw chip design, but they completely ignore the proprietary NVLink interconnect and Spectrum X networking that tie thousands of chips into a single giant brain. When millions of agents run simultaneously, moving data between chips is the ultimate bottleneck and Nvidia owns that networking standard. Second is Taiwan semiconductor. It does not matter whether the winning chip is designed by Nvidia, AMD or customuilt inside Google or Meta. Every advanced piece of silicon on Earth must be etched in TSMC's foundaries and packaged with their proprietary advanced packaging technology. Third is Broadcom. As open models make intelligence cheap, massive tech giants are designing custom application specific chips to lower their operational bills. Broadcom is the premier partner that co-designs those custom chips while simultaneously dominating the high-speed Ethernet switching silicon that connects enterprise data centers. Fourth is advanced micro devices. AMD is not just an alternative compute provider. Their server processors handle the massive system orchestration required to manage autonomous agents. AMD raised its long-term server market growth forecast from 18% up to 35% a year, adding 60 billion to their projections precisely while open models were expanding. Fifth is Micron Technology. Generative intelligence cannot operate without massive memory bandwidth. High bandwidth memory is mechanically brutal to manufacture and the entire global production capacity across the industry is effectively locked up and sold out for quarters in advance. Sixth is Veritive Holdings. You cannot run 40 trillion tokens a day through highdensity servers without generating staggering amounts of heat. Traditional air conditioning cannot handle modern server racks that draw upwards of 100 kilowatts, making Vertiv's liquid cooling and power distribution systems an absolute physical requirement. Seventh is Palunteer technologies. Cheap models are completely useless to an enterprise unless they can be integrated securely into messy corporate data without making things up. Palanteer's ontology provides the enterprise operating system that turns cheap commoditized intelligence into real measurable corporate cash flow. Let's go over the bull and bear cases. To hold these companies with absolute conviction, you have to look at the bare thesis dead in the eye. The primary risk is that Javon's paradox assumes consumer and enterprise demand is completely elastic. If there is an eventual ceiling on how much intelligence the global economy can actually use, collapsing model costs will eventually cut total capital expenditures instead of expanding them. The second real danger is margin compression across the hardware stack. If enterprises shift their workloads to cheaper open-source models running on commoditized infrastructure, chip designers may lose their pricing leverage. When hardware margins get squeezed from 75% down toward historical averages, valuation multiples can compress rapidly. You also have to remember that multi-trillion dollar market projections often come directly from the executive teams selling the equipment. If actual business adoption lags behind the massive physical capacity being built, the industry could face a cyclical hangover where data centers sit half empty. Those risks are real and they must be watched closely every single quarter. But betting against computing efficiency has been a losing trade for 70 consecutive years. When unit costs fall, utility expands and capital follows the bottleneck. When you step away from the daily noise and examine the underlying balance sheets, the fundamental reality becomes undeniable. The premier infrastructure providers in this space are not debt-ridden speculative startups burning capital on user acquisition. They are generating cash at an industrial scale, posting gross margins well north of 50% and converting an enormous share of every revenue dollar straight into free cash flow. Their operating leverage is structurally protected by multi-billion dollar manufacturing plants, decades of proprietary software, and insurmountable physical patents. You cannot spin up a competitor to an advanced semiconductor foundry or an enterprise liquid cooling network with a handful of venture capital. These businesses represent the physical choke points of the modern global economy. There are plenty of speculative software tickers trading in the market that offer flashier headlines and theoretical upside. But those assets carry existential downside if their underlying technology gets commoditized by the next open-source release. Allocating capital to the core infrastructure bottlenecks is the equivalent of betting on the house. Different software players will rise and fall. Open models will compete with closed ones and trading sentiment will swing wildly. But over a long enough time horizon, the house always collects its cut. We started with the core reality that cheap open artificial intelligence is out of the box and it's reshaping corporate spending from the ground up. You now have the economic blueprint of Javon's paradox, showing why 90% drops in compute cost trigger exponential demand for total compute capacity. You also have the historical context of that January sell-off, proving that market panic frequently misinterprets structural hardware expansion as an existential threat. Understanding that thesis on paper is only about 10% of the battle. The hard part is sitting at your desk when the opening bell rings, watching red candles flash across your screen, and executing your strategy without flinching. Having a strong macro argument means nothing if you lack the mechanical discipline and the framework to manage position sizes through deep pullbacks. Most retail investors try to navigate these generational technological shifts completely isolated. They bounce between contradictory media headlines, trade on raw emotion, and end up selling their highest conviction holdings right before the next upward leg of the cycle begins. Winning over a 5 to 10ear horizon requires an actual operating framework. It demands an environment where you can pressure test your long-term thesis against real market data alongside investors who view market drawdowns as strategic opportunities rather than reasons to panic. The mainstream crowd will keep debating which software model holds the top spot on the benchmark leaderboards this week. They will keep agonizing over whether training costs are dropping too fast, completely missing the title wave of token generation hitting the physical world. That leaves you with a much deeper, more consequential problem to solve. If token volumes expand by another 10x from here, the choke point will not even be the software algorithms themselves. It shifts straight to the physical world. electrical grid connections, high voltage transformers, and thermal management. Knowing which companies own the silicon is one thing, but knowing how to manage that exposure when the market throws another panic attack is what separates traders from generational investors. The structural bottlenecks are clearly visible. The data is on the table, and the only question left is how you position your capital before the rest of the market catches up. That separation between trading a headline and holding a generational asset is exactly why I built my Patreon community. Having the right macro thesis on tech infrastructure means nothing if you lack the framework to manage risk, size your positions, and hold the line when the market panics. Inside, I share every trade I take and the exact thinking behind it. You get to watch a thesis develop in real time from the initial buy through the ads, trims, and exits. Members are using this environment to build absolute conviction like AirTractor who recently booked over $5,700 on wheat and precious metals and over $4,400 on co mining just by having the discipline to trim and rebalance. If you are ready to stop reacting to the noise and start investing with serious intent alongside people who think long-term, the link is down in the description. If you made it this far, drop infrastructure in the comments. tribe check. That is the standard for investors who buy the bottleneck instead of the hype. For those of you who stuck around to the very end, I have a special reward strictly for the investors looking at the secondary ripples of what we covered today. Every talking head is obsessing over graphics cards and server racks. But almost nobody is looking at the hard physical ceiling of the entire buildout. That hard ceiling is pure raw electrical power. A modern highdensity data center cluster consumes more steadystate electricity than an entire midsize city. The lead times to order industrial high voltage step- down transformers have stretched past 3 to four years and utility interconnection cues are backed up across the entire country. This means the next explosive wave of capital is flowing straight to the companies that generate base load electrons and build the physical grid. On the generation side, look at independent power producers like Constellation Energy, ticker CEG, and Vistra, ticker VST. These companies own massive fleets of nuclear and gas plants capable of providing dedicated round-the-clock power directly to hypers scale data centers without touching the public grid. On the physical grid equipment side, look at GE Vernova, ticker GEV, and Eaton Corporation, ticker ETN. GE Vernova builds the heavy duty gas turbines and grid electrification gear that utilities desperately need to expand capacity. Eaton practically owns the industrial switch gear, circuit breakers, and power distribution hardware required inside the data center walls to keep those high voltage racks from melting down. When you realize that artificial intelligence is simply a machine that converts raw megawatts into high-v value digital tokens, your entire investment map expands. The software models may be free to download, but the power generators and electrical grid manufacturers will be printing cash for the next decade.
Commentaires 0
Connectez-vous pour rejoindre la discussion.
Se connecterAucun commentaire pour l'instant. Soyez le premier à partager votre avis !