Big Tech Is Moving On From Nvidia VIEW IN BROWSER 
Remember when you discovered a video game cheat code that basically let you win on autopilot? That was AI investing from 2022 to today. Up, up, down, down, left, right = Buy Nvidia (NVDA), layer in Microsoft (MSFT), Amazon (AMZN), and Alphabet (GOOGL), maybe sprinkle in Super Micro (SMCI) or CoreWeave (CRWV). Boom – infinite lives, exponential gains. You couldn’t lose. Nvidia alone is up more than 1,100% since the start of 2023. Just sit back, relax, and watch the money print itself. But what happens in every game? Eventually, the developers patch the exploit. And right now – while retail investors are still mashing the same buttons, expecting the same results – the hyperscalers are rewriting the source code. They’re done paying Nvidia’s 75% gross margins when they can build chips themselves for a fraction of the cost. And starting in 2026, they’ll flip the semiconductor industry on its head with custom silicon designed in-house. We’re calling this shift ‘The Great AI Decoupling.’ And if you’re not prepared, the portfolio you built during the easy-mode era is about to get obliterated. Here’s what’s coming… Why Custom Silicon Is Replacing Nvidia GPUs For the last few years, companies like Microsoft, Alphabet, and Meta (META) have been in a compute land grab. They needed GPUs yesterday, and price was no object. In fact, in 2025 alone, as data journalist Felix Richter noted, “Meta, Alphabet, Amazon and Microsoft are expected to spend between $350- and $400 billion in capital expenditure,” most of it dedicated to the AI buildout. But that math is breaking down. Running a massive, specialized AI model on a general-purpose Nvidia GPU is like using a Ferrari to buy groceries. Sure, it works – but you’re paying for a twin-turbo V8 when all you need is trunk space and decent gas mileage. Hyperscalers have realized that if they design their own chips – Application-Specific Integrated Circuits (ASICs) like Google’s Tensor Processing Units (TPUs) – they can optimize for their exact workloads and slash costs by 30- to 50% per inference operation. And this transition is happening right now. - Alphabet uses TPU v6 for a substantial portion of its internal AI training.
- Amazon just launched Trainium2 chips claimed to deliver up to 30% better price-performance than comparable Nvidia GPUs – and AWS is now pitching them hard to customers like Anthropic and Databricks.
- Microsoft has begun deploying its custom Maia AI accelerators in Azure datacenters and is integrating them into its cloud infrastructure to support large-scale AI workloads, including services that run models from partners such as OpenAI.
- Meta is in advanced talks to purchase billions of dollars worth of Google TPUs to reduce its Nvidia dependency.
In other words, the AI Boom’s ‘infinite budget’ phase is dead. The ‘efficiency’ phase has begun. And in an efficiency war, the generalist always loses to the specialist. | Recommended Link | | | | Nvidia helped trigger a historic AI boom that minted 600,000 millionaires in America. It’s now working on a new tech that could be even bigger than AI. According to Bank of America, “this [technology] could be the biggest revolution for humanity since discovering fire.” Click here to get the details. |  | | Four Custom Silicon Stocks to Buy for 2026 So, if $100-plus billion is shifting away from Nvidia and into custom silicon, where does it land? With the Enablers – the companies that sell the blueprints, the connectivity, and the lasers that make custom chips possible. Those are the stocks you want to own in 2026. And we’ve zeroed in on four plays that are particularly well-positioned to profit… Broadcom (AVGO): The Pick-and-Shovel Play for Custom AI Chips - The Pitch: If Google is the gold miner, Broadcom is who’s selling the pickaxes.
- Why It Wins: Google and Meta can’t build custom chips alone – they need Broadcom’s intellectual property. Broadcom provides the critical SerDes (Serializer/Deserializer) technology that moves data on and off chips at high speeds, plus the physical chip design architecture. Without Broadcom, there’s no TPU. Without Broadcom, there’s no custom AI chip at scale.
- The Catalyst: Broadcom just signed a massive deal with OpenAI to “jointly build and deploy 10 gigawatts of custom artificial intelligence accelerators as part of a broader effort across the industry to scale AI infrastructure.” This is the template: every hyperscaler building custom silicon needs Broadcom’s IP. CEO Hock Tan has said the company’s AI-related revenue could hit $60 billion annually by 2027. Broadcom isn’t just riding the custom silicon wave – it’s collecting rent on every chip that gets made.
Credo Technology (CRDO): The Cable King of AI Networking - The Pitch: The “Cable King” of AI networking.
- The Hidden Gem: Custom AI clusters run on standard Ethernet networking – but at extreme speeds (800 Gigabits per second, soon 1.6 Terabits), traditional copper cables can’t handle the signal. The data literally degrades after a few feet.
- Why It Wins: Credo makes Active Electrical Cables (AECs) – copper cables with embedded signal-boosting chips that extend range and reliability at ultra-high speeds. And they’ve got a near-monopoly on the tech. Exhibit A: Elon Musk’s Colossus supercomputer in Memphis – one of the world’s largest AI training clusters – runs almost entirely on Credo cables, not Nvidia’s. When xAI needed to connect 100,000 GPUs, they called Credo.
- The Trade: Credo is a small-cap with big volatility – but also explosive upside. The company’s revenue grew 272% year-over-year in its most recent quarter, and management sees Ethernet-based AI networking as a multi-billion-dollar TAM. If custom silicon becomes the standard, Credo could 10x from here.
Lumentum (LITE): Why AI Clusters Need Laser Technology - The Pitch: Light is faster than electricity.
- Why It Wins: As custom AI clusters scale into the tens of thousands of chips, copper cables hit a physical wall. You need fiber optics – and fiber needs lasers. Lumentum manufactures the electro-absorption modulated lasers (EMLs) that power the optical transceivers inside Google and Amazon’s datacenters. No Lumentum lasers, no long-distance, high-speed connectivity between chips.
- The Catalyst: The industry is upgrading to 1.6 Terabit Ethernet networking in 2025-2026, which requires next-generation EML lasers. Lumentum is among the leading vendors in this space and is deeply embedded with the hyperscalers. As custom silicon clusters expand, Lumentum’s revenue should scale proportionally.
Arm Holdings (ARM): The Royalty Machine Behind Every Custom Chip - The Pitch: The DNA of every custom chip.
- Why It Wins: When Microsoft builds its "Cobalt" CPU or Amazon builds its “Graviton” chip, they’re not inventing the underlying architecture from scratch – they’re licensing it from Arm. Arm’s instruction set is the foundation for nearly every custom CPU in the cloud.
- The Economics: Arm collects a royalty on every chip shipped – typically 1-2% of the chip’s selling price. As hyperscalers manufacture tens of millions of custom CPUs to pair with their AI accelerators, Arm’s royalty stream grows automatically. No extra R&D costs. No scaling challenges. Pure leverage. It’s one of the highest-margin business models in semiconductors.
The Final Word Wall Street is pricing Nvidia as if its dominance will last forever. It won’t. The capex budgets for 2026 are already being written, and they heavily favor custom silicon. The playbook for this is simple: - Don’t get trapped in yesterday’s trade. Crowded AI leaders can still run – but the risk/reward is changing as the market starts to look past the current bottlenecks. Trim your exposure to the “Nvidia Complex” (Nvidia, Oracle, CoreWeave, etc).
- Follow the money into America’s reinvestment wave. The next super-cycle is forming in the domestic, contract-driven “picks-and-shovels” ecosystem tied to this industrial reboot. Accumulate the “Custom Silicon Supply Chain” (Broadcom, Arista, Credo).
- Watch for the turning point. When the market sees that the old winners can’t keep dominating forever, leadership will shift fast. If Nvidia’s gross margins dip below 72% in their next earnings report, it is the first crack in the dam.
The AI revolution isn’t over. It’s just growing up. The “dumb money” is still chasing the GPU shortage. The “smart money” is building the factory that makes the GPUs obsolete. And the “smartest money”? They’re following the roadmap laid out by the federal government. Washington is systematically targeting chokepoints in the AI supply chain currently controlled by foreign interests – and backing U.S. companies to solve them with domestic solutions. And the stocks that win that federal backing will likely be among next year’s biggest winners on Wall Street. That’s why I’ve spent the past several months building a list – not of companies that already received government backing… but those most likely to be next… Because when Washington opens the money spigot and starts picking winners, early investors don’t just make money – they build fortunes. Get in front of the government’s next move before it’s announced. Sincerely, |
No comments:
Post a Comment
Keep a civil tongue.