Been tracking something interesting in the semiconductor space that most retail investors are probably sleeping on. The whole AI infrastructure buildout isn't slowing down—if anything, it's accelerating into 2026. And there's a clear play here for anyone paying attention to where the real money flows.

So here's what's happening. The five biggest hyperscalers and cloud companies in the US are about to dump somewhere between 660 to 690 billion dollars into AI infrastructure this year. That's nearly double what they spent last year at 380 billion. Add in what pure-play AI companies like OpenAI and Anthropic are throwing at infrastructure projects, and you're looking at an absolutely massive wave of capital deployment. This kind of spending creates opportunities if you know where to look.

The classic "picks and shovels" angle applies perfectly here. You don't need to bet on which AI company wins—you can just focus on the hardware makers supplying the infrastructure. Two names that have been on my radar are Micron Technology and Jabil. Let me break down why these could be serious winners.

Micron's situation is pretty straightforward once you understand the bottleneck. AI data centers are memory-hungry in a way that standard servers just aren't. Micron makes the DRAM chips that power these systems, specifically high-bandwidth memory designed for AI accelerators. These chips let GPUs and AI processors pull massive datasets quickly when training models or running inference.

Here's the kicker—back in 2019, Micron predicted that AI servers would need six times the DRAM of regular servers. They nailed that forecast. Now you've got companies like Nvidia packing increasingly dense memory configurations into their chips. Their new Vera Rubin processors launching this year are going to have close to 300 gigabytes of HBM. That's up from around 200 GB in their Blackwell B200 chips from last year. The memory requirements just keep climbing.

The supply situation is tight. Demand from AI data centers has created a genuine shortage of memory chips, which means manufacturers can push prices higher. I saw a Bloomberg report showing DRAM prices jumped 75 percent from December into January alone, building on increases that started in Q4. That trend isn't reversing anytime soon because it takes time to build new production capacity. The supply constraints are real.

When you look at AI prediction models and analyst forecasts for Micron's earnings, the numbers are compelling. Consensus estimates show earnings potentially jumping more than 5x in just two fiscal years. That's the kind of growth trajectory that tends to drive stock appreciation. The valuation math works too—Micron's trading at 24 times trailing earnings and 12 times forward earnings. Compare that to the Nasdaq-100 sitting at 31 trailing and 25 forward. You're getting a discount on a company positioned in the center of an AI infrastructure boom.

Now, Jabil is the less obvious play, but maybe that makes it more interesting. Most people don't immediately think of Jabil when they think AI infrastructure, but the company's been quietly positioning itself as a critical supplier to data centers. They handle manufacturing, engineering, design, and supply chain work across multiple sectors, but AI has become their primary growth driver.

Jabil builds the actual server racks, liquid-cooled systems, and power solutions that data centers need. They're not just riding the wave—they're actively investing to capture more of it. Last June they announced a 500 million dollar commitment to expand manufacturing capacity specifically for AI data center infrastructure. That's real capital deployment, not just talk.

The results are showing up in their guidance. Jabil just raised their forecast for AI revenue in fiscal 2026 to 12.1 billion with 35 percent growth expected. That's actually better than the 25 percent growth they were guiding for back in September. They also bumped operating margin guidance up by 10 basis points to 5.7 percent. When a company raises guidance twice in six months, that usually means visibility is improving.

Analysts are forecasting double-digit percentage earnings growth for Jabil going forward. The AI server market itself is expected to grow at 34 percent annualized through 2030, so there's a long runway here. Valuation-wise, Jabil's at 19 times forward earnings, which is a discount to the Nasdaq-100. The stock's already up 33 percent over the past three months, but if you look at the fundamentals and the market growth rates, there's still room for upside.

The broader picture is that AI infrastructure spending isn't a one-year phenomenon. It's a multi-year buildout, and the companies supplying the hardware are positioned to benefit significantly. Whether it's memory chips or server infrastructure, demand is outpacing supply right now, which creates pricing power and margin expansion opportunities.

Obviously do your own research and consider your own risk tolerance. But from an AI prediction and market forecast perspective, the picks and shovels trade in infrastructure hardware looks like it still has legs heading into the second half of 2026. Both Micron and Jabil seem worth having on your watchlist if you're thinking about exposure to this theme.