HScreen · ARM

ARM · Arm Holdings plc American Depositary Shares 5d old

conviction3

coverage tierT2-PARTIAL

horizon12 months

Return distribution probability-weighted scenarios, never collapsed to a point

-35%30%

+12%40%

+55%30%

EV +10.80%

why these probabilities: BASE (+12%, p=40): consensus delivered. Street FY2027 (ending 2027-03-31) revenue growth +23.5% and EPS growth +27.1% across 20/17 estimates, FY2028 +35.1%/+36.0% across 22/22 (source_id fmp:/stable/analyst-estimates?symbol=ARM&period=annual, as_of 2026-08-19, grade EST). Base assumes ~23-24% FY2027 revenue growth DELIVERED and the FY2028 ramp still credible at year-end, with EV/sales compressing modestly from 32x toward high-20s as the lower-margin silicon line begins to mix in (~30%+ GM, ARM/ARM_2026Q4.json:42). +27% EPS on a ~12% multiple de-rate ≈ +12%. This is DESCRIPTIVE — it is what is already priced, and I am not attempting to beat consensus with no model, no management access and no channel checks. BULL (+55%, p=30): revenue growth ~30%+ in FY2027 versus consensus 23.5%, driven by the AGI /own-silicon leg landing ABOVE the USD 2bn management is publicly working toward — which management has explicitly said it internally exceeds and is holding back only until supply capacity is visible (ARM/ARM_2027Q1.json:25), with demand already stated to be above the USD 90m/USD 910m FY2027-28 split (ARM/ARM_2026Q4.json:43). Add royalty rate continuing to offset handset unit weakness via v9/CSS (ARM/ARM_2027Q1.json:18). On that path FY2028 consensus gets revised UP rather than met, and a 32x sales multiple holds instead of compressing. ~+35% revenue with multiple held ≈ +55%. BEAR (-

not built — options-implied distribution on the same axis, with the KL number and where the disagreement sits
needs: desk-options implied bands (v1.4 Part 4a) — activates at 60% coverage; currently 33%

Visibility how far ahead this name can be seen

not built — visibility score
needs: visibility_score.py

Trust how much weight this name's own record can carry (credibility Z)

not built — credibility
needs: credibility.py

Sizing survival-first size (half-Kelly) vs the enforced ladder

kelly not computable — no data

Payout durability does the distribution survive a downturn?

PASS

DPS trajectoryinsufficient_history

interest cover nowx

interest cover at FORWARD rates42.45x

Falsifiers what would prove this wrong

The AGI/own-silicon ramp delivers at or above the guided USD 910m in FY2028, i.e. supply constraints are resolved rather than deferred
Royalty revenue per device continues to rise via v9/CSS migration, holding royalty growth positive despite flat-to-declining smartphone units
First-generation own-silicon gross margin lands at 30%+ as indicated, not below
No volume hyperscaler or major OEM CPU programme migrates away from the Arm ISA to RISC-V or an internal architecture within the horizon
Opex discipline holds — the +18% y/y R&D step-up does not accelerate further without matching revenue

Conflicts surfaced, never netted out (T4)

Street consensus target 341.82 implies +25.2% from 251.515; our probability-weighted expectation is +6.8%. Consensus rating is Buy with 19 buy / 6 hold / 2 sell across 27 analysts. We adopt the street's BASE case verbati	high
p_bear 30% versus node-pooled base rate 6.4% at -35% (23.6pp above) and historical 0.2% (29.8pp above). Defended in deviation_defence on three grounds: effective_n 2 and a post-2023-IPO sample containing no cycle; siblin	high
The revision tape is UNAVAILABLE: eps_drift and rev_drift both return n_up 0, n_down 0, direction flat, median_move_pct null, velocity 0.0 over a 365-day window, and outliers is empty — despite 362 estimate rows and 27 c	high
Consensus revenue dispersion widens sharply in exactly the years the valuation depends on: 1.2% for FY2026 (n=25), 8.4% for FY2027 (n=20), 17.7% for FY2028 (n=22), 13.8% FY2029-31. The FY2028 point estimate of +35.1% rev	medium
Management states no 11th-hour risk that a single unsecured component tips the silicon programme over, expressing confidence in partners (ARM/ARM_2027Q1.json:48), while in the prior quarter framing the FY2027-28 revenue	medium
Customer capex evidence is strong (AMZN backlog growing substantially and fully reflected in capex projections, AMZN/AMZN_2026Q2.json:43; GOOGL USD 514bn cloud backlog now including TPU system sales, GOOGL/GOOGL_2026Q2.j	medium
ARM shipping its own silicon (ARM/ARM_2027Q1.json:35) puts it in the same product category as the licensees who pay its royalties. The pack contains no evidence on how licensees have responded, and no evidence that they	medium

The case against independent adversary, different model family (P6) weakened

Version history from the research record

archived versions3

archived	hash	bytes
2026-08-17T14:27:28	186cd47113f1	37,115
2026-08-17T15:40:00	00c3a786bace	37,267
2026-08-19T08:32:01	d56bf28eec5b	35,264

What changed in the latest version

-	as_of: 2026-08-17
+	as_of: 2026-08-19
-	provenance:
-	producer: analyst-aisemi dossier run
-	generated: 2026-08-17
-	inputs:
-	- evidence pack ARM 2026-08-17T14:16:22Z (build_evidence_pack.py)
-	- state file /Volumes/SSD/Hermes/system/state-files/ARM.json (state_file_build.py 2026-08-17T13:17:28Z)
-	- estimates_query.py drift/outliers 2026-08-17T14:17:18Z
-	depot_classes_used: A,B only — Class D (own prior research) not read; blind re-score preserved
-	value: 'Arm Holdings plc (NASDAQ ADS, GB domicile, SoftBank-controlled) licenses CPU instruction-set
-	architecture and increasingly complete subsystems (CSS) to semiconductor makers and OEMs, earning
-	upfront licence fees plus per-unit royalties. FY2026 (to 2026-03-31) revenue USD 4.92bn, operating
-	income USD 908m, net income USD 904m, reported EPS 0.85. The business is mid-transition on two axes:
…	diff truncated

Rendered from canonical artifacts only (T1) — no terminal database exists. Every figure is read from the system's own files at render time; where an artifact is missing the panel names the dependency rather than showing an empty box. No execution path exists anywhere in this interface (T5), not even a link.