$Advanced Micro Devices(AMD)$ When I look at AMD, the main reason I'm optimistic is their strong positioning in the data-center and industry-wide AI build-out. The financials matter, but the real backbone of the thesis is the hardware they're shipping and the roadmap they're executing.

On the CPU side, AMD's EPYC line is doing the heavy lifting. Genoa (Zen 4) goes up to 96 cores, Bergamo (Zen 4c) pushes that to 128 cores for dense cloud environments, and Siena covers the lower-power edge and telco space. The next generation, Turin (Zen 5), is expected to raise the bar again on performance and efficiency. EPYC keeps taking share from Intel due to better performance per watt, higher core density, and stronger memory bandwidth. As AI workloads get more complex and require more CPU coordination, high-core-count CPUs become even more critical, which plays directly into AMD's strengths.

On the GPU side, AMD is finally gaining real traction as the only practical second source to NVIDIA for large-scale AI deployments. The MI300 family is central to that push. MI300A combines CPU and GPU in one package for HPC and hybrid AI workloads, while MI300X is a pure GPU accelerator with a large amount of HBM3 memory for training and inference. MI325X improves bandwidth and compute, and the upcoming MI350 series (CDNA 4) is expected to deliver a significant jump in AI performance. There's also MI450 further out, and early customer interest, according to the last earnings call, is reportedly stronger than expected. These accelerators are already being used in AI training clusters, inference systems, and supercomputers, especially when paired with EPYC CPUs.

AMD is in a very unique position now regarding the thinking around CPU and GPU ratio. If the optimal ratio approximates 1:2 or 1:3, then AMD is structured perfectly for this next phase. They have the expertise on both platforms to create an ideal blend.

The markets AMD competes in are expanding rapidly. Server CPUs are projected to grow into a $120–130 billion market by 2030, and AI accelerators could exceed $400 billion. If AMD keeps gaining share in both areas, the long-term earnings potential is substantial. They've also secured additional advanced-node capacity at TSMC, which is a big deal because AI chips are supply-constrained. This ensures manufacturing access so potential sales can be realized.

Another factor is where AI is heading. The next phase—more agentic, multi-step, real-time AI systems—leans heavily on CPU throughput and memory bandwidth, not just raw GPU compute. That shift aligns well with AMD's architecture and product mix. Analysts have been raising their price targets based on AMD's data-center momentum and AI roadmap. Even those cautious about valuation acknowledge the strength of the product lineup and the long-term opportunity.

The main risk is valuation. A lot of optimism is already priced in, especially with the CPU momentum in recent weeks. The stock depends heavily on AI demand continuing. I don't see anything immediate that would disrupt that demand. The ROI in AI isn't fully realized yet, but it's being figured out. I suspect a quiet awakening is starting regarding productivity gains and workforce changes. But from a hardware and positioning standpoint, AMD has several strong tailwinds behind it.

On a personal level, having experimented with AI over the past year, there's no doubt my productivity would have increased significantly if I were still working. Just thinking about the hours spent figuring things out in Excel versus now telling an AI agent what to do—that alone is a massive time saver. So the ROI feels quite real, and it's hard to imagine what the next 3-4 years will bring. We are witnessing the next industrial revolution, and it's going to progress at a pace far faster than anything we've seen in history.