The app for independent voices

5 stocks benefiting the most from AI β€” from GPUs to memory to infrastructure

Every bottleneck is becoming a profit center. Here are 5 names πŸ‘‡

$NBIS β€” Nebius Group

Pure-play AI cloud built around GPU clusters, not legacy workloads. Runs NVIDIA H100/H200 infrastructure across Europe and the U.S., with 2+ gigawatts of contracted power targeting 3GW by 2026.

Exited 2025 with $1.2B ARR, up from sub-$200M ~18 months earlier. Q4 operating cash flow reached $834M, driven by prepaid GPU contracts.

Capex hit $4.07B annually. Strategy is simple: lock power + GPUs early, pre-sell capacity, expand margins per deployed cluster.

$AAOI β€” Applied Optoelectronics

Every AI cluster depends on optical connectivity. AAOI builds it end-to-end. Owns laser design, semiconductor growth, and module assembly. Few competitors match vertical integration.

Secured a $200M 1.6T transceiver order for late 2026 delivery. Forward models point to $378M monthly transceiver revenue by Q2 2027, driven by 800G and 1.6T demand.

Also developing pump lasers for co-packaged optics. Next-gen architecture. Still early.

$IREN β€” IREN Limited

From Bitcoin mining to AI infrastructure. Same power, different workload. Controls data centers, energy, and compute. Eliminates middle margins.

Secured $3.6B GPU financing below 6% interest, with $1.9B Microsoft prepayment covering most capex. Already has $2.3B ARR contracted, targeting $3.4B by 2026.

Building a 1.6GW Oklahoma campus. Bitcoin roots turned into a power moat. Few can replicate at this scale.

$MU β€” Micron Technology

HBM is the bottleneck. Micron sits directly on it. One of only three companies globally producing high-bandwidth memory at scale.

Q2 operating cash flow reached $11.9B in a single quarter. Cloud memory unit alone delivered $7.7B revenue at ~74% margin.

HBM4 ramp ties directly into next-gen NVIDIA systems. When supply is constrained, pricing power shifts fast. Micron is already there.

$SNDK β€” SanDisk

Flash storage is no longer passive. It’s core AI infrastructure. Every dataset, checkpoint, and inference cache runs on high-speed NAND.

Q2 operating cash flow reached $1.02B, with gross margin expanding to 50.9% from 26.9% prior quarter. Guidance implies 65–67% margins as enterprise SSD mix increases.

PCIe Gen5 drives now qualified at multiple hyperscalers. Working on dedicated AI storage systems. Demand already exceeding supply.

Apr 9
at
12:24 PM
Relevant people

Log in or sign up

Join the most interesting and insightful discussions.