CALLARMabout 2 months ago7.6Kviews

Arm: the hardware wins the benchmark and loses the deployment

Framework parity is still a year and a half out for the workloads that actually pay. Hardware wins the evaluation and loses the rollout, and that shows up in guided revenue first.

Tomer Avraham@tomer_avraham

Evidence · 6 cards

The case

Arm: the hardware wins the benchmark and loses the deployment

Framework parity is still a year and a half out for the workloads that actually pay. Hardware wins the evaluation and loses the rollout, and that shows up in guided revenue first.

The edge

The Street says

Consensus models the cycle turning in the second half
Sell-side target clusters within 8% of spotAuto

I say

The turn already happened in order intake two quarters ago
My number is 14% above the street on the out yearCreator est.

Path to target

Entry: 131.62Auto
Out-year EPS: +18% vs consensusCreator est.
Exit multiple: 22x, in line with the last cycle peakCreator est.
= Target: 115.35Creator est.

Catalyst timeline

JUL 18, 2026 · PAST
Q results, margin guide reiterated
AUG 12, 2026 · PAST
Analyst day, capex framework updated
SEP 11, 2026
Next print, the number that decides this

Checklist

Order intake inflected
Pricing held through the renewal cycleAuto
Capacity expansion fundedCreator est.
Second source contract signed

The operative constraint is qualification time. A design win converts to revenue only after the customer's own validation cycle, which runs three to five quarters, so today's win is a fiscal-year-after-next revenue event. The gap shows in the deployment data rather than the benchmark data. Announced design wins have run well ahead of disclosed production deployments for six quarters, and the spread has widened rather than closed. The position: short on Arm, entry locked at publication and the exit dated rather than open-ended. The target is stated in the call block above and the horizon with it. Sized to the catalyst rather than to conviction, with the invalidation written into the kill-switch card instead of left implicit. The risk is that software parity arrives faster than the historical base rate. Sustained investment has closed gaps like this before, and if it closes here the deployment discount disappears with it.

Discussion · 5

Pauline VidalJUL 22

Finally someone put numbers on this. The utilisation assumption is where I disagree. Does the thesis survive a soft guide next quarter?

Yael BrennerJUL 22

Been long since the last one. The point about conversion rather than demand is the right one. What would you need to see to add rather than hold?

Ines FerreiraJUL 20

Not convinced, but well argued. The exclusivity window is the part the market keeps mispricing. Has anything in the last print changed the entry you would take?

Sam WhitfieldJUL 21

Good to see an invalidation level actually stated. The capital intensity point is the one I keep coming back to. Where does this break if rates back up another 50bp?

Sam WhitfieldJUL 20

Been long since the last one. The depreciation schedule really is the whole disagreement. What would you need to see to add rather than hold?