The Hardware Behind the Buildout

The “Elon Musk’s 10X Project” pitch from Brownstone Research is really two teases stacked together. The first resolves to a memory-chip maker, Micron Technology (MU). The second promises a “Next NVIDIA,” a chip with “50 times the computing power of the H100” that ran “179 times faster than a supercomputer.” Our cornerstone teardown walks through how both resolve and what each claim is worth.

AI chip stocks are the hardware layer of the AI buildout, and understanding the category means seeing how the accelerators, the memory, and the incumbent share of the market all fit together.

The Accelerator Market

The AI accelerator market is dominated by NVIDIA, which still controls roughly 80% of AI-accelerator share. That dominance has survived years of challengers because NVIDIA pairs its hardware with a software ecosystem, CUDA, that developers have built on for over a decade. Benchmarks matter less than that installed base, which is why faster competitors have repeatedly failed to take meaningful share.

The pitch’s “Next NVIDIA” tease resolves to AMD, whose MI-series accelerators do post strong numbers in at least some benchmarks. The “50 times the H100” and “179 times faster than a supercomputer” figures are isolated spec-sheet cherry-picks. A chip that wins one benchmark is not the same as a chip that displaces an incumbent with 80% share and a software moat.

Where Memory Fits

The first tease in the pitch is memory, and it is the more honest of the two. Every AI accelerator consumes fast memory, and each new chip generation carries more of it. The pitch argues that full self-driving needs “500 trillion operations per second” and that this is “only possible” with DRAM, dynamic random-access memory. The number is theater, but the direction is correct: memory, especially HBM, high-bandwidth memory, has become one of the tightest links in the AI hardware chain.

That is why the memory reveal matters. Micron is the only major US-based maker of DRAM and HBM, vertically integrated, and one of a three-way oligopoly alongside SK Hynix and Samsung. The HBM market is on track to reach roughly $33 billion by 2027, and the shift from HBM3e to HBM4 is where the margin gains sit.

The Two Claims, Side by Side

The pitch’s first claim, that memory is an AI chokepoint, is well sourced. The second claim, that a “Next NVIDIA” is about to unseat the incumbent, is not. NVIDIA’s roughly 80% share has not been meaningfully dented by AMD or anyone else, and the software moat is the reason. That contrast is worth holding in mind when the pitch tries to blend the two stories into a single thesis.

The “silent partner” framing on the memory side has the same problem. Memory is a three-way market, and after the pitch was recorded it was widely reported that Tesla had asked both SK Hynix and Samsung to develop HBM4 prototypes for its AI chips. There is no exclusive supplier, just as there is no imminent dethroning of NVIDIA.

How to Think About AI Chip Stocks

AI chip stocks span accelerators, memory, and everything in between, and they do not move as one. The accelerator side is a winner-take-most market with a dominant incumbent. The memory side is a three-way oligopoly with a cyclical pricing history and a genuine HBM transition underway.

For a closer look at the memory name at the center of the tease, see our Micron stock article. For the Tesla-chip angle the pitch leans on, see our Tesla chips analysis. For background on the firm behind the pitch, see our Brownstone Research profile.

The Scoreboard and the Cycle

The teaser’s headline number, +911%, compares a $92.50 tease price to a current price near $935.39, and it is backward-looking. It records the gain someone earned who bought when the pitch was first recorded, not the return available to a reader seeing the ad today. Micron’s market cap is roughly $1.06 trillion, and the stock trades around 15 times forward adjusted earnings.

The memory side is also cyclical. DRAM has swung between shortage and oversupply for decades, and all three suppliers have a history of ramping capacity at once and then pricing themselves into a downturn. The HBM buildout is real, but it unfolds inside that cycle, which is the main reason to weigh the memory story on its own rather than through the pitch’s urgency. The promo’s countdown, a promised “conference call around April 23,” has already passed with no supplier reveal.

The fair way to approach AI chip stocks is to separate the accelerator story from the memory story, check each on its own merits, and treat benchmark claims with the skepticism they deserve.

Ready to see the research? Click here to access the Near Future Report.

NewsletterVetter is an independent publication. We receive compensation from some of the services we review through affiliate links. Nothing on this site is investment advice. Always do your own research.