Skip to main content
All posts

Why a 112G SerDes Designer Is the Hardest Analog Hire in Semiconductor

By Game 7 Staff5 min read
Abstract 112G PAM4 eye diagram showing four noisy signal levels with narrow eye openings and realistic jitter

112G/224G SerDes designers are the scarcest analog talent in semiconductor. Why they're so hard to hire, and how to find one who's silicon-proven.

When a chip program slips, the post-mortem usually points at verification coverage or a timing-closure crunch. On any design that has to move data fast, there is a quieter constraint that gets set much earlier, by one of the smallest teams on the project: the analog engineers who own the high-speed serial links. Among them, the SerDes designer is the single hardest analog hire in the market right now, and the gap is widening.

If you are staffing a program with a high-speed interface roadmap, this is the role most likely to be on your critical path and least likely to be sitting on a job board. Here is why, and what actually separates a SerDes designer who can carry your link from one who can only execute inside someone else's architecture.

The Link Is Where Bandwidth Lives

Every generation of compute pushes the serial link harder. PCIe 6.0 moved the standard to PAM4 signaling at 64 GT/s per lane, the 800G Ethernet standard (IEEE 802.3df) was approved in 2024, and the OIF's CEI-224G framework now defines the electrical path beyond 112G per lane. Each of those steps is not an incremental tweak. Moving from NRZ to PAM4, then doubling the symbol rate again, changes the analog problem in kind, not degree.

When the link rate doubles, signal-to-noise margin collapses, channel loss climbs, and the equalization that recovers the signal gets dramatically more demanding. That is the work. And it lands almost entirely on the analog designer.

What a SerDes Designer Actually Owns

A SerDes designer builds the transistor-level transceivers that connect chips to the outside world: PCIe, Ethernet, CXL, die-to-die. The block list looks deceptively short, but each item is its own discipline. Continuous-time linear equalization (CTLE), decision-feedback and feed-forward equalization (DFE/FFE), the clock-and-data-recovery loop, the PLL that clocks it, and the jitter budget that ties them together against a lossy channel.

This is done at the schematic level in Cadence Virtuoso and simulated in Spectre against a foundry PDK, with the layout hand-drawn because analog routing parasitics are part of the design, not an afterthought. A SerDes engineer thinks in volts, picoseconds, and dB of insertion loss, not logic gates. They are closer to a physicist than a software engineer, and that is exactly why the role does not scale the way digital does.

Why You Can't Synthesize Your Way Out of It

Digital design has synthesis: write RTL, let the tool map it to gates. Analog has no equivalent. Every transistor is hand-sized, hand-biased, and hand-placed, and the judgment behind those choices is deeply experiential. Most principal analog designers have 15 to 20 years in the craft, because the feedback loop, from a design decision to silicon that proves it right or wrong, is measured in tape-outs, not sprints.

The pipeline feeding this skill is thin. University EE programs have shifted toward digital and software for two decades, so the supply of new analog designers is structurally small while demand from data center and AI compute is structurally large. That mismatch is why SerDes rates sit at the top of the analog band and why the search takes longer than any other discipline.

The Node Tax on Analog

Advanced process nodes make the problem harder, not easier, for analog. TSMC's N2 process and its gate-all-around (GAA) nanosheet transistors change device behavior in ways that matter for analog performance: matching, parasitics, and the headroom a designer has to work with. A SerDes designer who has silicon-proven a link at an advanced node, and can explain how the device physics shaped their choices, is a fundamentally scarcer hire than one who has only worked at mature nodes.

Senior vs. Principal: The Distinction That Matters in a Screen

A senior SerDes designer can own a defined block within an established link architecture and deliver it to spec. A principal designer architects the link budget itself: chooses the equalization scheme, partitions the jitter budget, sets the targets the rest of the team designs against, and has silicon-proven results they can correlate from bench measurement back to simulation when first silicon does not behave.

The fastest way to tell them apart is to ask about a link that did not close on the first pass. A principal designer will walk you through how they isolated whether the problem was in the channel, the equalization, or the CDR, and what they changed. That story is hard to fake, and it is the signal you are hiring for.

What to Look for on a Resume

Specificity is the filter. The candidates worth a conversation name their protocols and rates (PCIe Gen5/Gen6, 112G PAM4), reference eye diagrams and bathtub curves as routine, list tape-outs by process node, and can speak to the channel they designed against. Vague claims of "high-speed experience" without a rate, a protocol, or a node are the tell that you are looking at adjacency, not depth.

Why the Shortage Won't Ease Soon

This is not a cyclical blip. Deloitte projects the global semiconductor industry will need more than one million additional skilled workers by 2030, and the Semiconductor Industry Association estimates a U.S. shortfall of roughly 67,000 workers over the same window. Analog, and high-speed SerDes specifically, sits at the scarce end of that gap because the skill cannot be trained quickly or automated. If a high-speed link is on your roadmap, the time to start the search is before the schedule depends on it.

How Game 7 Finds Silicon-Proven SerDes Talent

Game 7 Staffing recruits primarily in semiconductor, hardware, and platform-software disciplines, which means our SerDes and analog network is built on technical conversations, not keyword searches. We can tell a link architect from a block-level contributor, and we surface engineers who are mid-program or between engagements, not the ones browsing job boards.

If you have a high-speed interface on your roadmap, submit your hiring needs and we'll put a short list of silicon-proven designers in front of you. If you are an analog or SerDes engineer, share your background with us and we'll get back to you, learn about your goals and career projection, and get you placed where you want, doing what you want.

FAQ

Frequently Asked Questions

What does a SerDes design engineer do?

A SerDes (serializer/deserializer) designer builds the transistor-level high-speed serial transceivers that move data between chips, including PCIe, Ethernet, and CXL links. The work covers equalization (CTLE, DFE, FFE), clock-and-data recovery, and PLLs, designed in tools like Cadence Virtuoso and Spectre and proven in silicon against channel-loss and jitter budgets.

Why are 112G and 224G SerDes designers so hard to find?

Analog design cannot be synthesized the way digital can; every transistor is hand-sized and hand-placed, and the learning curve runs 15 to 20 years. Demand from data center and AI compute has outpaced a structurally thin pipeline, so silicon-proven high-speed SerDes designers are among the scarcest hires in the industry.

What's the difference between a senior and a principal SerDes designer?

A senior designer executes a defined block within a set link architecture. A principal architects the link budget, chooses the equalization scheme, partitions the jitter budget, and has silicon-proven designs at advanced nodes that they can correlate from bench measurement back to simulation.

Can SerDes and analog roles be filled on contract?

Yes. Principal-level analog and SerDes engineers are regularly placed on contract when a program needs high-speed link expertise for a specific tape-out window, without carrying that rare skill set as permanent headcount.

Written by

Game 7 Staff