Cadence rides agentic AI to a record backlog and its highest-ever quarterly raise — while the tape sleeps in a 23% drawdown

Q2 2026: 24% growth, 40%+ IP acceleration, Intel finally in the fold; the stock still sits well below its June peak.

CDNS · Earnings Call · 2026-07-27

The agentic inflection

Something genuinely new is showing up in Cadence's Q2 call. The company has been talking about Agentic AI for several quarters — it was the top keyword in Q1 too — but this quarter the emphasis shifted from "early traction" to real monetization. “We exited the quarter with record backlogs that was above our expectations.” — Anirudh Devgan, President and Chief Executive Officer · 2026-07-27 That record stands at $8.1B, and management was unambiguous about what drove it: “Cadence is now the only provider with agentic solutions spanning the full electronic system design flow” — Anirudh Devgan, President and Chief Executive Officer · 2026-07-27 — from digital/analog design to verification, packaging, and PCB.

The raise was the biggest single-quarter revenue raise the company has ever made, to 19% growth for the year — and the story is broad-based, not dependent on one product. Anirudh Devgan: “This is the highest we have raised annual revenue in a single quarter.” — Anirudh Devgan, President and Chief Executive Officer · 2026-07-27 New super agents are landing: AuraStack for PCB/packaging (up to 15x productivity), ChipStack with 20+ customer engagements (40x faster RTL validation), ViraStack for analog, InnoStack for advanced-node SoC. The Al agent is no longer a roadmap item — it's a demand accelerator that repeatedly invokes the physical EDA tool base. This is exactly the logic Anirudh laid out a year ago:

the way this improvement will happen… will be through this three-layer cake. We will have agents like we have super agents. Our tools are central, will continue to be central to that.
Anirudh Devgan · 2026-02-17

The "three-layer cake" — agents on top, physics-based middle layer engines, compute at the bottom — is now producing numbers, not just narrative.

Intel: the 20-year dog finally wagging

For years, Intel was Cadence's known weak spot — a 10-to-20-year-old problem, as Anirudh put it. The multi-year 14A collaboration announced this quarter (alongside deepened Samsung and Rapidus ties) marks a genuine strategic pivot, not just an incremental win. “We are actually pleased with our discussions with Intel… we'll have more to say on our engagement with Intel.” — Anirudh Devgan · 2026-04-27 That was Q1; now there's an actual contract. The IP business grew over 40% year-over-year, with memory IP (HBM, LPDDR6, UCIe) and advanced packaging in white-hot demand as AI inferencing drives varied memory architectures. Notably, the Intel arrangement is incremental on top of existing business, and the foundry base is diversifying — Intel 14A, Samsung 2nm, Rapidus — each bringing new IP and EDA content. John Wall's guidance this quarter is strategically important:

For our updated outlook for 2026, we now expect revenue in the range of $6,260 to 6,340 million… with the useful assumption that export control regulations that exist today remain substantially similar.
John Wall, Senior Vice President and Chief Financial Officer · 2026-07-27

The export-control caveat is a quiet acknowledgment of the one macro risk that could poke a hole in this story.

Fundamentals confirm — almost

The numbers back the narrative. Total Revenue hit $1,584M in Q2, up 24% year-over-year, with every product group growing double digits. Operating Margin rose to 45.5% non-GAAP, though management flagged deliberate second-half investment around Intel and Hexagon D&E integration. The one yellow flag is Free Cash Flow (less SBC), which fell to $169M from $286M a year ago — likely billings timing around the backlog build, but it's the one number that doesn't yet confirm the acceleration.

The tape disconnect

Here's the rub: the stock is still 23.4% below its June 2, 2026 peak of $416.39, even after this blowout quarter. The recent 90-day shape shows a violent round trip — up 56% over six weeks, then down 23% over the following ten. The fundamentals and the guidance clearly improved; the price tells you the market is either unconvinced the agentic AI TAM is real, worried about the export-control assumption, or simply de-rating a name that ran too far too fast. For an investor who believes the "three-layer cake" is finally baking, the drawdown is the entry; for a skeptic, it's confirmation that even a record backlog can't outrun a lofty price-to-revenue multiple of Price to Revenue of 13.9x. The next two quarters — and the durability of that $8.1B backlog converting to FCF — will decide which camp is right.