Seagate: AI Data-Center Build-Out and Multi-Year Storage Runway

Created by@hypertonxvia MCP
July 28, 2026 at 9:52 PM

Seagate: AI Data-Center Build-Out and Multi-Year Storage Runway

Executive view

Seagate does not see the AI data-center build-out approaching a cyclical peak. Management describes a multi-year infrastructure expansion with unusually concrete storage-demand visibility through 2028 and customer planning extending into 2029 and beyond.

What Seagate is seeing now

  • Data-center demand represents approximately 90% of Seagate's exabyte shipments.
  • Cloud nearline demand has recorded three years of sequential quarterly exabyte growth, with management reporting no evidence of a slowdown.
  • The supply-demand gap has widened versus several quarters ago.
  • Contracts define product configurations and pricing across calendar 2027.
  • The vast majority of nearline exabytes are already allocated into calendar 2028.
  • Customers are seeking to extend planning horizons through 2029 and beyond rather than pulling them back.
This is strong evidence of actual procurement commitments, not only optimistic hyperscaler forecasts.

Why management expects the cycle to last

Seagate sees AI storage developing in successive waves:
  1. Training: Large historical and synthetic datasets must be created and retained.
  2. Inference and agents: Persistent context and KV-cache data increase the need to store and reuse information rather than recompute it.
  3. Physical AI: Robots, autonomous vehicles, and sensor networks generate large volumes of unstructured video that return to cloud or edge infrastructure for repeated learning.
Management characterizes agentic and physical-AI storage demand as being in the very early stages. Its core thesis is that AI makes data cumulative: applications create, retain, and reuse more information over time.

Important caveat

Seagate cannot yet quantify the eventual HDD attachment rate from agentic AI; management said it could vary significantly by application. The company's 2028 commitments also include traditional cloud and video workloads, so they should not be viewed as purely AI-related. That diversification may nevertheless make the storage outlook more durable.

Bottom line

Seagate's evidence supports a strong build-out through 2028, tangible planning into 2029, and potential additional demand into the 2030s from agents and physical AI. The contracted horizon is more concrete than the longer-term application thesis.

Source

Seagate Technology Holdings Q4 2026 earnings transcript, July 28, 2026.
Show actions

Start a new chat seeded with this artifact.

Built with Modeledge MCP

Connect your MCP client to research filings and earnings calls, then build a research note like this one.

Connect your agent