AlphaLens
Research
财报15小时前 · Morgan Stanley

Lotes Co. Ltd: Asia AI Summit Feedback; 2Q GM Under Pressure but Improvement in 2H

3533.TW
中文EN⚠ quality lint: translated(zh): ZH 文章混入英文词 33 个 (ratio 0.02)

Lotes Co. Ltd: Asia AI Summit Feedback; 2Q GM Under Pressure but Improvement in 2H

Core Conclusion

Lotes’ 2Q gross margin faces a transitory headwind from plastic resin cost increases, but a steep recovery from 2H is probable as cost pass‑through contracts activate. Server socket volumes — the company’s main value driver — are forecast to grow 15‑20% YoY in CY26, with a preliminary +15% YoY indication for CY27, supported by AMD Venice and Intel OakStream ramp schedules. Desktop deterioration is a secondary negative that tempers near‑term sentiment but does not alter the structural AI‑server growth thesis. The investment case turns on 2H margin normalization and server programme execution; ARM socket adoption remains a significant but low‑probability downside risk in the current forecast period.

What the Market May Be Underestimating

  • Margin rebound timing: Plastic prices began rising only in March, so 2Q is the single quarter that fully captures the raw‑material spike. 2H should see contracts re‑priced, allowing gross margin to reverse sharply. Consensus may be extrapolating 2Q pressure into 2H.
  • Server resilience versus desktop weakness: Desktop is now guided down to a 10‑15% YoY decline (previously -5‑10%), yet server growth of +15‑20% YoY more than offsets the absolute revenue impact, given Lotes’ mix shift toward higher‑ASP server sockets. The market may be overweighting the consumer PC miss.

Evidence Chain

  1. Plastic cost drag is precisely timed: “Price of plastics started to rise only in March, with a full quarter of reflection in 2Q, but GM will kick in from 2H.” This suggests a 1‑2 quarter lagged pass‑through mechanism, typical for Lotes, and implies a V‑shaped margin trajectory.
  2. Server unit growth rerated upward: The CY26 volume forecast of +15‑20% YoY was reiterated with a further +15% preliminary guide for CY27. This two‑year visibility is uncommon and reduces the uncertainty around the AI buildout cycle.
  3. New socket ramps provide H2 catalysts: AMD’s Venice Socket is expected to ramp from 2H26; Intel’s OakStream in CY27. Both are full‑slot solutions with higher content value, reinforcing the revenue‑ASP‑margin uplift narrative.
  4. ARM socket risk is acknowledged but unconfirmed: Only Ampere and Microsoft currently use a CPU socket. Lotes is in active discussions with other CSPs, “but there is no confirmation that CPU socket will be adopted.” This leaves a binary TAM expansion uncounted in consensus.
  5. Incremental NPO and SOCAMM opportunities: SOCAMM production begins in 3Q. Two NPO engagements exist — one via LINTES (outsourcing) and one direct — but no additional detail since the 1Q call, suggesting these remain early‑stage and not yet margin‑accretive.

Key Divergence and Risks

  • Plastic cost persistence: If resin prices stay elevated beyond 2Q and contract renegotiations fail, the expected 2H GM recovery may be shallower. Lotes has historically passed through raw‑material moves within 6 months, making this a low‑probability but high‑impact tail risk.
  • ARM socket non‑adoption: The strategic risk is that major CSPs designing proprietary ARM CPUs forgo a socket entirely, shrinking Lotes’ total addressable market. Current engagement with CSPs provides no assurance of conversion.
  • Desktop deterioration could worsen: The full‑year desktop decline of 10‑15% may prove optimistic if enterprise PC demand weakens further. However, this segment is a shrinking share of Lotes’ profit pool.
  • Execution on new programmes: Venice Socket and SOCAMM ramps carry production and yield risks that could delay the 2H margin inflection.

Valuation and Trade Implication

Lotes trades on a premium multiple rooted in its server socket quasi‑monopoly. The 2Q margin compression is likely a last‑quarter trough; shares that retreat on the miss could offer a tactical entry ahead of confirmed cost pass‑through and socket ramp announcements in late 3Q. The key trigger is management’s explicit 3Q margin guidance; a sequential recovery above 50‑100 bps would validate the thesis. Conversely, failure to recapture margin by 3Q would call the structural pricing power into question. The server volume trajectory supports a mid‑cycle multiple, but near‑term risk/reward hinges entirely on the GM recovery timeline. No explicit target price is set, but any dip driven by 2Q results without a downgrade to server guidance should be seen as an opportunity.