Agent-Built Database Substrate
é. urcades@edouard
−10.1%
since Sep 30
The idea
I believe agentic coding is about to make building custom, purpose-built databases cheap and routine, which commoditizes specialized database vendors and pushes the value down to what those engines run on. So this basket skips database companies and instead owns hyperscaler cloud, storage and memory hardware, and networking silicon, plus observability platforms that profit from monitoring a more fragmented database landscape regardless of which engines win. It holds large-cap cloud as ballast, keeps 11% cash to add on dips, and is sized for the next couple of years rather than a decade.
What it owns
- Agent-built custom databases will need somewhere to run and store their data, and these platforms provide the foundational compute and object storage that any bespoke engine would rent. · Amazon.comIts S3 storage and EC2 compute are the default substrate where most custom databases would run, making AWS a beneficiary regardless of which specialized engines get built on top., MicrosoftIt owns both the agent tools writing custom databases through GitHub Copilot and Azure as the cloud substrate where those engines would run and store data., OracleIts cloud infrastructure business grows as new workloads deploy to OCI, betting that Oracle's value comes from substrate capacity rather than its legacy database products., AlphabetGoogle Cloud provides storage, compute, and its own agent models as a substrate where custom databases could be deployed at scale.29%
- Performance-tuned custom databases drive demand for the specialized memory, flash, and disk components that sit beneath them, making these suppliers beneficiaries of infrastructure diversification. · Micron TechnologyPerformance-optimized custom databases drive demand for the high-bandwidth memory and DRAM that sit in data center infrastructure., SandiskCustom databases tuned for performance depend on fast NAND flash storage, making it a supplier of the critical memory component for specialized engines., Western DigitalCost-effective object storage at scale still relies on hard drives, positioning it as a supplier to hyperscalers building cheap bulk storage for expanding data volumes., Seagate TechnologyMost data still archives to low-cost bulk storage on high-capacity drives, making it a beneficiary of growing total data volume across specialized systems.20%
- More bespoke databases mean more data traveling between specialized engines and their storage, driving demand for the networking chips and switches that carry it through data centers. · Arista NetworksIts data center switches carry the traffic between bespoke engines and their storage, making it a beneficiary of growing networking demand in infrastructure., BroadcomIts networking chips sit in nearly every data path inside data centers, positioning it to benefit from increased data movement driven by specialized databases., Marvell TechnologyIt builds custom silicon and interconnects that move data inside cloud infrastructure, selling to hyperscalers that need specialized chips for data movement.15%
- Thousands of specialized database engines across different stacks create a monitoring nightmare, and these platforms collect revenue based on the complexity and fragmentation they help customers track. · DatadogMonitoring thousands of specialized databases is complex, and it collects fees for observability across fragmented infrastructure regardless of which engine wins., DynatraceIt provides automated monitoring for increasingly complex and custom infrastructure stacks, positioning itself as the observability layer for agent-driven deployments.13%
- Developers and agents will need places to write, test, deploy, and run custom databases close to users, and these platforms compete to be where that code gets built and served. · CloudflareIts edge platform, serverless functions, and distributed storage are designed for exactly the kind of lightweight, custom database deployments agents and developers would create., DigitalOceanIf custom database development becomes routine for smaller teams and solo developers, it provides affordable cloud infrastructure where that work would run., GitLabAgents and developers writing custom databases need platforms to version, test, and ship code, and it provides one of the primary software delivery systems for that workflow.13%
- Cash11%