Deep Research

Synopsys (SNPS)

4 min read

Snapshot

  • Ticker: SNPS (Nasdaq)
  • Bucket: Compute & AI Megacaps
  • Q1 2026 position: 136,925 shares, $54.3M, 1.09% of 13F
  • HQ: Sunnyvale, California
  • What it does: #1 electronic design automation (EDA) vendor plus semiconductor IP and, post-Ansys, simulation & analysis software — the toolchain every AI chip is designed with.

Business Overview

Synopsys sells the software that designs chips: EDA tools (synthesis, verification, hardware-assisted emulation), licensable Design IP (interface PHYs, controllers, foundation IP), and — since closing the ~$35B Ansys acquisition on July 17, 2025 — engineering simulation spanning electronics, structures, and fluids. The duopoly with Cadence is protected by decades of tool lock-in at every advanced-node design team.

FY2025 (ended Oct 2025) revenue was 6.13B, including a 2.28B, beating guidance, with EDA up 8%+ on hardware-assisted verification and advanced-node strength, Design IP at 652M. Non-GAAP EPS was 9.63–9.71B revenue and $14.72–14.80 non-GAAP EPS.

Financial Trajectory

PeriodRevenueYoYNotes
FY2024$6.13B~+15%Pre-Ansys
FY2025$7.05B+15%Ansys closed Jul 2025, contributed $757M; Q3 FY25 IP miss → ~36% drawdown
Q2 FY2026$2.28B—Non-GAAP op. margin 39.5%; EPS 9.63–9.71B

Why Atreides Owns It

Synopsys is a “wafers” derivative one level up the stack: every custom accelerator Baker talks about — TPU, Trainium, the Broadcom/Marvell ASIC wave, Nvidia’s own roadmap — is designed in Synopsys (or Cadence) tools and increasingly assembled from Synopsys interface IP (PCIe, UALink-adjacent SerDes, HBM PHYs). Proliferation of AI chip design starts, which Atreides’ whole book implies, is direct demand for EDA seats, emulation hardware (ZeBu/HAPS), and IP licenses, regardless of which chip wins. The Ansys merger extends the franchise into multiphysics simulation just as power and thermal become the binding constraints of AI silicon — a clean rhyme with “watts and wafers.”

But the fund has never trusted the price. Three entries and two full exits in six quarters — in at Q1 2025, out by Q2, back small in Q3 2025 (around the September crash), out again in Q4, and back with $54M in Q1 2026 — track a stock that whipsawed on China export rules, the IP-business stumble, and Ansys integration noise. The pattern suggests Atreides views SNPS as structurally right but episodically mispriced, renting it when the duopoly trades cheap relative to AI design-start momentum.

Position History

QuarterTypeShares/NotionalValue% of 13F
Q4 2024—not held——
Q1 2025Common152,488$65,394,4791.99%
Q2 2025—not held——
Q3 2025Common38,876$19,181,0300.37%
Q4 2025—not held——
Q1 2026Common136,925$54,288,0241.09%

Three separate entries (Q1 2025, Q3 2025, Q1 2026) and two complete exits — the most in-and-out name in the Compute bucket. The small Q3 2025 re-entry coincides with the post-earnings collapse; the Q1 2026 rebuild restores roughly the original position size at a similar dollar value but ~10% fewer shares. No options have appeared on the line.

Risks

  • China exposure: export-control regimes have already hit revenue once; China is a double-digit percentage of sales and a recurring policy football.
  • Ansys integration: a ~$35B merger of different sales motions and accounting (note the Q2 FY26 gross-channel restatement); synergy math is unproven.
  • IP business volatility: Design IP is lumpier and more customer-concentrated than EDA; it caused the 2025 stumble and is still down YoY.
  • AI writes RTL: the long-tail risk that AI-native design tools (including from cloud vendors or startups) erode seat-based EDA economics — ironic exposure for a fund that believes “anything you can verify, you can automate.”
  • Own-book signal: Atreides has exited this name twice in six quarters; it is demonstrably not a hold-through-drawdown position.

Sources

Backlinks

Graph View