Thoughts From the Road – Seagate Edition

Research evidence excerpt

Thoughts From the Road – Seagate Edition

hat range with no

expected step function going forward. As revenue grows, investment

grows modestly, primarily HAMR test equipment to maintain output, but

4–6% remains the right framework. On debt, ~$400M of convertible

notes are expected to be paid down over the next two quarters. The

longer-term goal is to reduce gross debt by $2–3B and potentially

become net-cash-positive in the not-so-distant future. Cash and free

cash flow generation is strong. On M&A, STX continues to look at

opportunities but nothing large is in scope. Management plans to

increase the dividend with significant cash available for buybacks.

HDD vs. NAND Architecture — A Crucial Distinction: Data center

architecture has not changed in the last 10–15 years: storage is HDD,

compute is NAND, and customers buy both as separate components.

Data resides on HDD, and when it needs to be run it moves to NAND.

Neoclouds currently only do compute, so they import data from a CSP,

run the application, and return the data (using mostly NAND). However,

if neoclouds eventually want to operate independently of the CSPs and

run compute for other end-customers, they will at some point need their

own storage tier (HDDs).

Other Incremental Points:

HDDs Will Not Be the Gating Technology: No new data center has

been built without HDDs, and the industry has been able to serve all

new data centers and refresh cycles to date. Near term — even in a

less-than-ideal supply/demand environment — the industry can keep

up. Looking further out, demand exceeds the industry's ability to

produce on an "unconstrained" basis (i.e., assuming no other

bottlenecks like power, DRAM, semis, or cables). In reality, some other

component will almost always be the binding constraint, which means