When I invest in AI, I don’t chase the obvious names — I focus on the picks and shovels. The hardware and core infrastructure AI simply can’t run without. These are the 🔑 manufactures you would find if you ever visited a data centre & the companies powering AI & LLMs ⚡️Power & Infrastructure $IREN — AI compute powered by energy-backed datacenter infrastructure $VST— Electricity generation fueling hyperscale AI training demand $VRT — Cooling and power systems for dense GPU clusters $CCJ /$CCO— Nuclear fuel enabling always-on AI datacenter electricity ☁️Compute + Cloud $NBIS— Neo-cloud renting GPUs for enterprise AI training workloads $GOOG — Global cloud + proprietary models monetizing AI compute $AMZN — Largest AI infrastructure rental platform via hyperscale cloud 💾Semiconductors $AMD — Accelerators competing to power large AI model training $ASML — Lithography machines required to manufacture advanced AI chips 🛜Networking $ANET — High-speed networking connecting thousands of GPUs into clusters 🔐Data + SaaS + Security $SNOW — Structured enterprise data feeding AI models and analytics $NET — Edge network delivering and protecting AI applications globally $RBRK — Backup and cyber-recovery protecting enterprise AI datasets Phase 3 of the AI trade isn’t chips. It’s the infrastructure that powers them. I’m Sure there’s many other great stocks out there , these are just the stocks I own, I can’t own every single stock. Always do your own research 🙏🏽
36K views
34 Comments
Kap @kap100 · 6moEdited
@bdinvesting This is good list but There are key missing pieces to make it a "Complete" AI Ecosystem and very imp in 2026. The "Brains" (Foundation Models) $MSFT : Through their partnership with OpenAI and their own "MAI" models, they remain the primary gatekeeper for enterprise AI. Meta ($META ): Their open-source Llama models have become the industry standard for companies building custom AI without being locked into a single cloud provider. Anthropic and OpenAI >> Upcoming IPOs The "Enablers" (Manufacturing & Specialty Tech) You have $ASML , but the physical creation of every chip on your list depends on one company not listed: TSMC ($TSM ): They manufacture nearly 100% of the advanced AI chips for NVIDIA, AMD, and Apple. Broadcom ($AVGO ): While you have Arista ($ANET ) for networking, Broadcom owns the custom "ASIC" market (building custom AI chips for Google and Meta) and the physical switches that move data within the clusters. The "Adopters" (Vertical AI & Edge) This is where the real revenue growth is shifting in 2026; companies using AI to solve specific problems. The Edge: Apple ($AAPL ) or Qualcomm ($QCOM ). As AI moves from massive data centers to your phone (On-Device AI), these companies capture the consumer side. Physical AI/Robotics: Tesla ($TSLA ) or NVIDIA ($NVDA ). NVIDIA is no longer just a "chip" company; in 2026, they are the platform for "Physical AI" (robotics and autonomous factories). HBM (High-Bandwidth Memory): This is the most critical component. It is physically stacked on top of or next to GPUs like those from NVIDIA or AMD. Without HBM3E or HBM4, these chips effectively "starve" for data. Key Player: Micron ($$MU ). Your current picks (like $NBIS , $VRT , and $IREN ) are very "Infrastructure-heavy," which is a smart play for the 2026 energy and compute crunch. However, if the market shifts from "building the tech" to "using the tech," this might find missing the massive gains in the software and edge sectors.
Jonti LM@ragnaros · 6mo
NVT- Liquid cooling + electrical applications NRG- Utilities just purchased 19 GW gas powered EU- US domestic uranium supply scaling up APH- electrical connections and wiring for data centres
