Full-Stack AI Buildout
brad artziniega@bradartziniega
−5.6%
since Sep 30
The idea
I believe AI keeps expanding through the whole economy over the next year or two, so I built this to own the buildout itself rather than slow adopters. It spans chips and hardware, hyperscalers, power and grid, data center infrastructure, and a couple of AI software winners, concentrated in my highest-conviction pick at each layer rather than spread thin. I keep a small cash reserve on hand to buy dips, and I'm holding this with a multiyear mindset as the stack matures.
What it owns
- Every layer of AI infrastructure runs on processors and memory, so these components get paid first in the buildout. · NVIDIAIt supplies the processors that train and run most AI systems across the industry, giving it first claim on buildout spending., Taiwan SemiconductorIt manufactures the advanced semiconductor nodes where nearly every leading AI chip, from multiple vendors, gets produced., BroadcomIt manufactures custom AI chips and networking silicon for the biggest cloud companies designing their own processors., Advanced Micro DevicesIt offers a credible alternative AI processor to competitors and is expanding upward into AI model development., Micron TechnologyIt manufactures high-bandwidth memory that every advanced AI processor requires, with few alternative suppliers globally.31%
- The giant cloud companies are spending the capital on AI infrastructure and collecting returns by monetizing that capacity. · MicrosoftIt is among the largest spenders on AI infrastructure buildout and monetizes that capacity through Azure cloud services and enterprise AI products., AmazonIts AWS cloud platform hosts significant AI workloads, and it manufactures its own processors for those services., AlphabetIt controls a complete stack spanning custom chips, data centers, AI models, and monetization through search and advertising., OracleIt has contracted to host AI model training at scale, with a large backlog of committed revenue from that capacity., Meta PlatformsIt is investing heavily in AI infrastructure and already converting that spending into measurable returns through improved advertising performance.24%
- Data centers consume far more electricity than the grid was built to supply, and the companies upgrading power infrastructure are the critical bottleneck. · Vertiv HoldingsIt supplies cooling and power infrastructure that keeps high-density AI server clusters from overheating and failing., GE VernovaIt sells power generation and grid equipment that data centers need to meet electricity demands beyond what existing grids provide., Constellation EnergyIt operates the largest nuclear power fleet in the country, providing the always-on electricity that data centers require under long-term contracts., EatonIt manufactures electrical distribution and backup systems that connect power plants to server racks in data centers.20%
- Someone has to construct, network, and operate the physical facilities housing AI infrastructure, and that spending is accelerating. · Arista NetworksIt manufactures the network switches connecting AI chips within training clusters, where speed between processors is critical., Digital RealtyIt owns data center real estate where AI infrastructure is deployed, and leasing demand has outpaced available space., Dell TechnologiesIt assembles AI servers from components and is a top supplier to cloud companies and enterprises building AI infrastructure.11%
- These companies have already converted AI capability into growing revenue, not just development-stage products. · ServiceNowIt sells software that helps enterprises deploy AI agents and automation into their operations, capturing the adoption side of AI expansion., Palantir TechnologiesIt wins contracts to integrate and deploy AI systems within government and large enterprise customers at scale.6%
- Cash8%