원문 보기
I'm horrible at forecasting, but I'm very good at first principled reasoning of the size of the opportunity. We're going to need more fabs. However, uh remember that we're also improving the algorithms and the processing of it um tremendously over time. It's not as if the efficiency of computing is what it is today and therefore the demand is this much.
In the meantime, I'm improving computing by a million times every 10 years while demand is going up by a trillion times. And that has to offset each other. Does that make sense? And then there's technology diffusion and so on so forth. That's just a matter of time. But it doesn't change the fact that one day all of the computers in the world will be changed 100%.
every single data center will be all of those general purpose computing data centers 100% of the trillion dollars worth of infrastructure will be completely changed and then there'll be new infrastructure built on even on top of that while the mainstream media obsesses over AI bubbles and ROI timelines a structural tectonic shift is quietly occurring beneath the surface of the global economy and it is worth trillions most investors are looking at the current surge in capital expenditure as a temporary spike but Nvidia IA CEO Jensen Hang is signaling something far more permanent.
He is arguing that we aren't just building new data centers. We are witnessing the systematic obsolescence and total replacement of the world's entire $1 trillion legacy computing footprint. For 30 years, the world built retrievalbased infrastructure, giant digital libraries of pre-recorded files waiting to be fetched by CPUs.
Today, Huang posits that this model is stern dead. The future is generative, where intelligence is manufactured in fast, real time by accelerated GPUs. This isn't just a minor upgrade. It's a weighty wholesale transition where 100% of generalpurpose data centers must be gutted and rebuilt into AI factories.
The numbers backing this are staggering. Hyperscaler capex from the big five is projected to exceed $600 billion in 2026 alone, a 36% jump from 2025. While Hang is improving computing efficiency by a million times every decade, he notes that demand for generative intelligence is exploding by a trillion times.
This massive gap creates an infinite demand loop for specialized hardware, networking, and power systems. We are moving from a world that retrieves information to one that produces tokens of intelligence as a utility. In this video, we are looking at the three perfect stocks that sit at the intersection of this $1 trillion replacement cycle.
Let's begin with the Titan at the center of this storm, Nvidia Corporation, ticker symbol NVDA. While the broader market fixates on quarterly fluctuations, the real story here is a massive supply constrained regime change. Nvidia has moved far beyond being a mere chip designer. It is now the official architect of a once- ina generation infrastructure overhaul.
As Hang points out, we are rebuilding the world's data centers to manufacture intelligence. This has shifted Nvidia's business model from a cyclical hardware vendor to a structural infrastructure provider, evidenced by its elite gross margins holding steady near 75%. A figure unheard of in traditional semiconductor down cycles.
The numbers backing this bullish thesis are undeniable. Hypers scale capital expenditure is projected to hit a staggering $600 billion by 2026 with 75% of that flow directed toward GPUs and AI networking. While critics point to deployment bottlenecks as a risk, savvy investors recognize these as evidence of overwhelming demand.
We are seeing something fundamentally abnormal in the tech world. Legacy hardware like H100 GPUs is actually appreciating in value rather than depreciating. This scarcity regime confirms that Nvidia is the gatekeeper of a constrained asset. Furthermore, the transition to full rack solutions like the Blackwell NVL72 represents a massive shift in average selling price and operating leverage.
Each rack isn't just a server upgrade. It's a million-doll factory that generates tokens of intelligence. Looking ahead to the upcoming Q4 earnings on February 25th, the setup is asymmetric. While the stock has traded sideways in a consolidation band between $150 and $200, earnings estimates have continued to climb.
This has compressed Nvidia's forward multiple to roughly 18 to 19 times, well below its 5-year average of 45 times. Effectively, the market is pricing in a slowdown that isn't showing up in the data. With the expansion of Nvidia AI enterprise software and price hikes on high-end GPUs, the company is set to prove that its moat is built on more than just silicon.
It is built on the CUDA ecosystem that $4.5 million developers now call home. For the active investor, this isn't just about a beat and raise. It's about owning the engine of the $1 trillion replacement cycle. To keep these trillion dollar engines running at peak performance, you need the essential neural pathways provided by our second stock, Broadcom, ticker symbol AVGO.
While the market has been distracted by a SAS apocalypse narrative, fearing that AI vibe coding will disrupt traditional software, the reality is that Broadcom owns the AI trifecta, worldclass networking, custom XPU compute engines, and a high margin infrastructure software mode through VMware that is virtually untouchable.
As thousands of GPUs are clustered into these new AI factories, the networking input/output becomes the ultimate bottleneck. Recent data shows that networking constraints can leave high-cost GPUs idle up to 35% of the time. Broadcom is the only player with the hardware software stack to solve that idle time crisis, making their technology a non-negotiable requirement for hyperscalers like Meta and Google.
The financial setup here is a classic buy the dip opportunity. Despite Meta and Google raising their 2026 capex midpoints to a combined $300 billion plus, Broadcom's stock has recently seen a 30% pullback, compressing its forward multiple to a very attractive 32 times. This valuation ignores the fact that Broadcom is a free cash flow machine, delivering 10% annual dividend growth while simultaneously slashing billions in debt.
We are looking at a company poised to drive over 50% growth this fiscal year with a 48% free cash flow margin. As the industry moves from training massive models to the inference phase, where AI is actually put to work, Broadcom's custom TPUs and XPUs are expected to gain significant share over more expensive generalpurpose GPUs.
As we head toward their important March 4th earnings report, the indicators are screaming double beat. The integration of VMware is already 90% complete among top tier customers, bringing in massive 93% gross margin license revenue that will likely offset any near-term supply chain costs. With a staggering 73 billion backlog and a leading role in the transition to open Ethernet standards, Broadcom isn't just riding the AI wave, it's building the surfboard.
We are targeting a medium-term price objective near $492 per share. As the market realizes that Broadcom software isn't being disrupted by AI, it is being empowered by it. While Nvidia and Broadcom build the high-end frontier, the sleeping giant finally waking up to claim the rest of the data center is advanced micro devices, ticker symbol AMD.
While everyone has been staring at Nvidia, AMD has been quietly engineering a $1 trillion infrastructure takeover that many active investors are completely missing. The inflection point isn't coming, it is already here. In its latest quarter, AMD's data center revenue surged 39% year-over-year to a record 5.4 billion, now accounting for more than half of its total sales.
This is no longer just a PC chip company. It has transformed into a compounding AI growth engine. The core of Jensen Huang's replacement thesis is that legacy infrastructure must be scutted and AMD is the primary alternative for hyperscalers looking to diversify away from a single supplier monopoly.
AMD's EPYC server CPUs are already cannibalizing Intel's market share, moving from nearly 0 to 30% of the market in record time. But the real gamecher is the agentic AI shift. As AI moves from simple chat bots to complex agents that run loops of tasks, the demand for high performance CPUs, not just GPUs, is exploding.
Lisa Sue projects the server CPU market to grow by strong double digits in 2026, positioning AMD to potentially capture 50% of that market by 2028. The financial setup is a smart gift to the disciplined investor. Following a recent post-earning sell-off triggered by skeptical short-term noise, the stock dropped nearly stern 20%.
Even as fundamentals reached record highs, AMD is currently generating a record wealthy $2.1 billion in free cash flow and holds a $10.5 billion cash horde. This allows them to confident self- finance the massive 2H 20226 ramp of the MI450 Instinct GPUs and the Helios rack scale systems without taking on a bold dime of new debt.
Management is targeting a staggering amazed 60% annual revenue CAGGR for the data center segment through 2028 with the MI450 series set to deliver 1.5 times the memory capacity of the competition and a important mega deal with OpenAI going live later this year. AMD is firm, derisked and ready for a massive rerating.
We are standing at the edge of the most significant infrastructure replacement in human history. The shift from retrieving data to generating intelligence is a trillion dollar mandate that cannot be ignored. Whether it's Nvidia's core processing power, Broadcom's essential networking glue, or AMD's aggressive expansion into the rest of the rack, the opportunity for investors is clear.
The $1 trillion shift is happening quietly, but the returns for those who move early will be anything but silent. Make sure to do your own due diligence, but keep these three powerhouses on your radar as the generative era begins in earnest.
Comments 0