To The Moon

1️⃣ Broadcom’s Position in the AI Race

Broadcom’s surge is justified — the reported partnership with OpenAI marks a major strategic win. The company’s strength lies in custom silicon and networking chips, not general-purpose GPUs like Nvidia’s. If OpenAI truly intends to co-develop application-specific AI accelerators, Broadcom could carve a niche similar to what TSMC is to fabrication — indispensable but behind the scenes. This makes it more of a picks-and-shovels play rather than a pure AI model bet. It’s unlikely to dethrone Nvidia, but it could easily become the next big infrastructure enabler after AMD, especially as hyperscalers look for cost-efficient, domain-specific chips.

2️⃣ On OpenAI’s “Crazy Galaxy”

OpenAI is evolving from a software lab into a full AI ecosystem architect — touching models, inference hardware, and now possibly semiconductors. This vertical expansion is reminiscent of Apple’s A-series and Tesla’s Dojo strategy. The move shows OpenAI’s long-term ambition to control the full AI stack, reduce reliance on Nvidia, and enhance efficiency for inference workloads. It’s aggressive, but strategically coherent.

3️⃣ Potential Future Picks

If OpenAI’s hardware vision expands, the next beneficiaries could include:

TSMC (TSM) – inevitable manufacturing partner for custom AI silicon.

Marvell (MRVL) – strong in networking and data-centre interconnects.

Micron (MU) and SK Hynix (000660.KS) – leaders in HBM memory essential for AI accelerators.

Cadence (CDNS) and Synopsys (SNPS) – EDA software providers that profit from every new chip design wave.

In short, Broadcom’s entry signals that the AI hardware landscape is broadening beyond Nvidia and AMD — and investors may want to start scouting the second layer of beneficiaries across memory, interconnect, and design software.