$NVDA For every 1 dollar earned, 50 cents flows into these 10 stocks:
1. TSMC $TSM
For every 1 dollar Nvidia earns, approximately 0.12–0.15 flows to TSMC.
The chips must be manufactured on TSMC’s production lines. Advanced processes and CoWoS are both bottlenecked here. Without TSMC, everything downstream comes to a halt. It is the most tightly tied in and the steadiest beneficiary.
2. Super Micro Computer $SMCI
For every 1 dollar earned, the complete-server stage accounts for approximately 0.08–0.12 in business, with Super Micro Computer taking a share.
Nvidia sells chips, but customers buy entire racks. Assembly, testing, and delivery all fall on server manufacturers. When chip sales rise, rack orders rise as well.
3. Micron $MU
For every 1 dollar earned, approximately 0.05–0.08 flows to Micron.
In a high-end GPU, HBM has become too expensive to ignore. Only three companies worldwide can produce it, and Micron takes a share. The stronger GPU sales are, the higher the memory bill.
4. Broadcom $AVGO
For every 1 dollar earned, approximately 0.04–0.06 flows to Broadcom through networking chips.
Turning thousands of GPUs into a single supercomputer requires switching and networking silicon. Customers often pay for it separately, but without this layer, Nvidia’s cards cannot be assembled into a cluster.
5. Vertiv $VRT
For every 1 dollar earned, approximately 0.03–0.04 flows to companies like Vertiv through power and liquid cooling.
The more expensive GPUs are, the more power they consume and the more they fear heat. Without power delivery and liquid cooling, a rack is just scrap metal. The more Nvidia ships, the more power and cooling each data center needs.
6. SanDisk $SNDK
For every 1 dollar earned, approximately 0.02–0.03 flows to SanDisk.
The more cards are sold, the more storage a cluster needs. Each individual order is not as large as those of TSMC or Micron, but it follows volume.
7. Marvell Technology $MRVL
For every 1 dollar earned, approximately 0.02–0.03 flows to Marvell Technology.
Once its design gets into a rack, it must be purchased again for every rack shipped afterward. Nvidia’s growth will compound downstream through interconnect chips.
8. Vistra $VST
For every 1 dollar earned, the power layer represents an estimated long-term bill of 0.02–0.04, with Vistra benefiting from power generation and data center electricity supply.
This money usually does not appear on Nvidia’s purchase orders. It goes toward electricity bills and power-plant investments. The more cards are sold, the tighter power becomes and the more valuable power generators become.
9. Credo $CRDO
For every 1 dollar earned, approximately 0.01–0.02 flows to Credo through high-speed connectivity.
No matter how powerful a GPU is, without connectivity it is just an isolated island. Every additional rack means another round of connectivity components sold. Each individual order is not large, but it is tightly linked to shipment volume.
10. Applied Optoelectronics $AAOI
For every 1 dollar Nvidia earns, approximately 0.01–0.02 flows to companies like Applied Optoelectronics through optical modules.
Moving data between GPUs relies on 800G optical modules. The more cards there are, the tighter the demand for optical ports. They earn from the cost of moving data within the cluster.
Copy Nvidia’s homework!
1. TSMC $TSM
For every 1 dollar Nvidia earns, approximately 0.12–0.15 flows to TSMC.
The chips must be manufactured on TSMC’s production lines. Advanced processes and CoWoS are both bottlenecked here. Without TSMC, everything downstream comes to a halt. It is the most tightly tied in and the steadiest beneficiary.
2. Super Micro Computer $SMCI
For every 1 dollar earned, the complete-server stage accounts for approximately 0.08–0.12 in business, with Super Micro Computer taking a share.
Nvidia sells chips, but customers buy entire racks. Assembly, testing, and delivery all fall on server manufacturers. When chip sales rise, rack orders rise as well.
3. Micron $MU
For every 1 dollar earned, approximately 0.05–0.08 flows to Micron.
In a high-end GPU, HBM has become too expensive to ignore. Only three companies worldwide can produce it, and Micron takes a share. The stronger GPU sales are, the higher the memory bill.
4. Broadcom $AVGO
For every 1 dollar earned, approximately 0.04–0.06 flows to Broadcom through networking chips.
Turning thousands of GPUs into a single supercomputer requires switching and networking silicon. Customers often pay for it separately, but without this layer, Nvidia’s cards cannot be assembled into a cluster.
5. Vertiv $VRT
For every 1 dollar earned, approximately 0.03–0.04 flows to companies like Vertiv through power and liquid cooling.
The more expensive GPUs are, the more power they consume and the more they fear heat. Without power delivery and liquid cooling, a rack is just scrap metal. The more Nvidia ships, the more power and cooling each data center needs.
6. SanDisk $SNDK
For every 1 dollar earned, approximately 0.02–0.03 flows to SanDisk.
The more cards are sold, the more storage a cluster needs. Each individual order is not as large as those of TSMC or Micron, but it follows volume.
7. Marvell Technology $MRVL
For every 1 dollar earned, approximately 0.02–0.03 flows to Marvell Technology.
Once its design gets into a rack, it must be purchased again for every rack shipped afterward. Nvidia’s growth will compound downstream through interconnect chips.
8. Vistra $VST
For every 1 dollar earned, the power layer represents an estimated long-term bill of 0.02–0.04, with Vistra benefiting from power generation and data center electricity supply.
This money usually does not appear on Nvidia’s purchase orders. It goes toward electricity bills and power-plant investments. The more cards are sold, the tighter power becomes and the more valuable power generators become.
9. Credo $CRDO
For every 1 dollar earned, approximately 0.01–0.02 flows to Credo through high-speed connectivity.
No matter how powerful a GPU is, without connectivity it is just an isolated island. Every additional rack means another round of connectivity components sold. Each individual order is not large, but it is tightly linked to shipment volume.
10. Applied Optoelectronics $AAOI
For every 1 dollar Nvidia earns, approximately 0.01–0.02 flows to companies like Applied Optoelectronics through optical modules.
Moving data between GPUs relies on 800G optical modules. The more cards there are, the tighter the demand for optical ports. They earn from the cost of moving data within the cluster.
Copy Nvidia’s homework!
0
