Synopsys (SNPS) posted a strong quarter with earnings that beat, driven by design adoption up 81%. Design closure runs 10 times faster for customers using its products. The company makes the software that designs chips, and more and more it designs the things built around chips.
Why the numbers jumped
Nvidia (NVDA) said its Vera Rubin chip is coming to market fast, with big design wins. Nvidia expects $40 billion in revenue for every gigawatt of compute installed, up from $18 billion with Blackwell. It designs these new chips so much faster because it uses Synopsys software. Other users include Cisco (CSCO), MediaTek, and Samsung Foundry, all reporting that 10 times faster design closure.
Synopsys is also lapping a large acquisition it made a year ago, and those results now show up in the top line, GAAP gross margins, and operating margins. The acquired company was Ansys (formerly ANSS), a big European firm with a large European customer base. That deal lets Synopsys extend its design work past the semiconductor itself. It already had products in that area, but Ansys works at a different level. The acquisition was hard to close and got caught up in China-US talks. Once it cleared and closed, the Synopsys sales force went out with a new product to sell.
CFO Sheila Glazer said the sales force is finding doors open, not only for the new Ansys products but for other Synopsys offerings too, because the new product gives the team momentum. The company also cut costs heavily during the merger, which lifted margins. That let it raise guidance for revenue, margins, and earnings for both the coming quarter and the coming year.
Synopsys had one of its worst days ever last September, tied partly to the acquisition moving slower than expected. Eleven months later the yearly losses aren't erased, but they've narrowed. The stock is down 25% on the year and had a strong day on these results. Did the quarter finally prove the turnaround is real? The numbers show it, and the market now recognizes how smart the Ansys deal was.
Hardware and software moving together
The IGV software index rose alongside chips on the day. These two pillars of the tech trade have swung back and forth like a ping-pong match, rewarding traders who were long hardware and short software or the reverse. That zero-sum pattern is fading as the two categories start to merge.
Many guests over recent months called the "SaaS apocalypse" silly and said software stocks shouldn't be down, but they are. I don't think it's silly at all. What AI can do is jaw-dropping. The success of Claude Code and OpenAI's Codex is going to replace a lot of products in the world. Investors have to do the homework and figure out what they own, because these aren't just wiggly lines on a screen.
Synopsys gives one answer to the key question: are there Synopsys offerings the world won't need if Claude can just figure it out? The company has a business called design IP. Think of it like Legos. If you're designing a chip and want it to connect to an optical connector, to memory, or to something else, those connections are hard to build from scratch, so Synopsys licenses them to you. The design IP business had been shrinking, which raised worry that AI tools were the cause. This quarter design IP rose 11% year over year and rose sequentially too, even after Synopsys sold a big IP portfolio to GlobalFoundries (GFS) in the last six months or so. That tells us Claude Code can't do everything, and that using Synopsys IP is a faster way to design semiconductors and the systems around them. Wall Street can exhale, because the stored IP clearly has value and is producing rising sales.
The 30-year moat
Making semiconductors is hard. Synopsys uses AI to make it easier for customers to reach the 30 years of design information it has gathered. No important semiconductor built in the last 30 years was designed without Synopsys software. That makes its product far better than any AI tool, because no AI tool has access to that data. Now AI lets designers work faster than ever.
This is one reason for a decoupling from the old cycles. Memory has been more cyclical than any other part of technology over 40 to 50 years, and there's debate over how long Micron (MU) can keep charging high prices. What's different now is longer-term innovation driven by OpenAI and Anthropic, which isn't tied to a 24-month product release cycle from Microsoft (MSFT) and Intel (INTC). The tech world used to be locked to that cycle. With tools like Synopsys and more chip companies designing GPUs and CPUs, including Nvidia (NVDA), AMD (AMD), and Intel (INTC), plus private firms going public such as Positron, innovation happens faster and pulls away from the historic cyclicality. That is part of why tech has had such a strong run.
Optical networking, Marvell, and capacity
Do we eventually return to the cycle, and how sustainable are these numbers? Marvell (MRVL) reports later in the day, and its story is a shifting portfolio mix toward the interconnect business. The focus has moved from the chip itself to how chips talk to each other. Optical networking connects these systems, and those connections keep getting longer, now linking data centers at the speed of light and linking servers at the speed of light using optical communication. Marvell (MRVL) is an important part of that.
It can't go on forever, but the better approach is a beginner's mind: look at what's actually there instead of what you remember from the past, because some things are different and some are the same. What we don't see is a slowdown. Things are picking up. Every data center company and every equipment provider says things aren't slowing. Everyone says they're limited by capacity, nothing else.
On the circular financing that Nvidia (NVDA) and others get criticized for, I haven't done much work on it and am only now starting, since many others are already digging in. If every company says capacity is the biggest problem, it makes sense for Nvidia to invest its balance sheet, whether in Corning (GLW), Lumentum (LITE), or Coherent (COHR), or in NeoClouds to give them a lower cost of capital so they can secure power and land. Nvidia's balance sheet is growing fast, with hundreds of billions in extra free cash flow expected over the next year or so. Putting some of that money to work to keep the buildout going is a reasonable use of the balance sheet. A lot of smart people disagree, but it looks like they're seeding future growth, and it seems to be working.
Positron is a private company I hold stock in that builds products using Synopsys software.
