Pick a Major With Your Eyes Open: Employment Odds and Starting Pay by Field

Prediction is very difficult, especially if it's about the future. - Neils Bohr

My post yesterday - Engineering Jobs, Class of 2026 and Beyond - sparked some questions. Students (class of 2026 high school and those in college that are early in a major selection) along with some anxious parents want to know not just how engineering was doing, but how other fields compared. So here are the actual numbers, by major, before you or your student commits four years and a tuition bill to a course of study. But.... before we get into it the data in this post will age. Check it before you choose and keep tracking it.

Two numbers matter when you pick a major: whether you can get a job at all, and whether that job actually requires your degree. The Federal Reserve Bank of New York tracks both for recent graduates. In early 2026, the overall unemployment rate for new college graduates sat at 5.7 percent. More telling is the underemployment rate, the share working in jobs that do not require a college degree: 41.5 percent. That means nearly four in ten new graduates are serving coffee or managing a retail floor regardless of what their diploma says. The gap between majors on this metric is enormous.

Criminal justice majors face a 67.2 percent underemployment rate. Performing arts is 63.2 percent. General business, the catch-all major, runs 52.8 percent, far higher than accounting at 17.9 percent or business analytics at 27.2 percent. On the other end, nursing sits at 9.7 percent underemployment and computer science at 16.5 percent, per St. Louis Fed data. Computer science unemployment in 2025 ran about 6 to 7 percent, higher than average, but underemployment was low: if you landed a job, it was likely a real one in the field. Unemployment and underemployment pull in opposite directions for some majors; you need both numbers.

On the salary side, NACE's Winter 2026 Salary Survey projects these starting averages for bachelor's graduates: computer sciences at $81,535, engineering at $81,198, math and sciences at $74,184, and business at $68,873. Social sciences are the only category where projected salaries dropped, down 1.7 percent from last year. The gap between the top and bottom is real: computer science and engineering graduates start about $35,000 to $40,000 ahead of education, fine arts, and sociology majors, who typically open around $42,000 to $48,000.

Salary rankings and employment odds do not always point the same direction. A college senior in May 2026 expected to earn roughly $80,000 one year out, according to a Clever survey. The actual median starting salary was closer to $60,000. The inflation in expectations is not random: students hear the computer science headline number and generalize. STEM majors do earn more, on average. But within STEM the spread is wide, and geography compresses those national averages fast. A business administration grad taking a first job in a smaller market should expect something closer to the lower end of the employer band, not the $68,873 national average.

Before committing to a major, look up three things. Check the BLS Occupational Employment and Wage Statistics for what similar roles pay in your actual market, not the national average. Ask your college career services office for its own first-destination data: school-specific placement rates beat any national survey for your situation. And check the New York Fed's outcomes-by-major table directly. 

Every graduating class generates the same mix of relief and dread, usually with the same absence of specifics. The numbers above will not guarantee anything, but they put the decision on better footing. Four years is a long time. So is forty years of a career built on a choice made without doing a little research.

Bohr is right about the future. Things are changing very fast. Keep running the numbers.

Engineering Jobs, Class of 2026 and Beyond

I've been watching the Class of 2026 posts on LinkedIn. 

Overall (all majors) grad hiring is up just 1.6% from last year, and 45% of employers rate the market only "fair," the weakest reading since 2021. Engineering is holding up better than most fields.

Average starting salary for engineering graduates is $81,198, up from $78,731 last year. Petroleum engineering leads individual majors at $100,750. Master's level averages $92,873. These are base salaries; total compensation runs higher in defense and energy.

Mechanical is the second most in-demand bachelor's degree this cycle. At least 60% of employers surveyed plan to hire mechanical engineers from the Class of 2026. Electrical is being pulled hard by data center construction, which is running at record pace across the country. Civil engineers are busy in Texas, Oklahoma, and Colorado. Job growth projections through 2034: 11% for industrial, 9.1% for mechanical, 7.2% for electrical. Median salary (the midpoint wage; half of workers in that field earn more, half earn less) for those three disciplines currently sits at $101,140, $102,320, and $111,910.

Projected job growth through 2034 and median salary by engineering discipline. Source: U.S. News / BLS.

Three open engineering positions exist for every qualified candidate right now. Nearly half of working U.S. engineers are 50 or older. Retirements are outpacing new graduates in several disciplines, and that gap is not closing anytime soon. Hiring demand for engineers is projected to grow 13% by 2031. Defense contractors and infrastructure firms are actively recruiting now, not waiting for fall.

GPA screening dropped from 73% of employers in 2019 to 42% today. Internships and real project work are what move resumes forward. Some AI and data analytics fluency helps, particularly in electrical and systems roles. You do not need to be a researcher; you need to show you can use the tools.

The shortage of qualified candidates is real. Not a bad time to finish an engineering degree.

The Pools at the Bottom

Pic courtesy of fws.gov

Tim and I were friends in elementary school and through high school. We fished and we hiked all over the Western Massachusetts  Berkshires. We specialized in finding waterfalls and the pools at the bottom that held brook trout. Small but beautiful. Tim always caught more than me. We almost always let them go, careful with how we handled them. He’s also the one who turned me on to Hemingway, including Big Two-Hearted River, both parts. A man goes fishing alone, and the whole story is what he doesn’t say.

High school ended. We went to different colleges. Jobs came, then families, and life pulled us in different directions. Forty years went by before we found our way back to each other.

This kind of drift happens to almost everyone. Nobody decides to lose a friend. It happens in small pieces: you mean to call, you don’t call, the next week comes and you don’t call then either. The years pass like that, fast, faster than you expect, until you realize you don’t actually know how someone you once knew well is doing anymore.

Judy, Tim’s wife, died this past April. Since then there have been a couple of text messages, nothing more. I haven’t picked up the phone. I never met his two sons. Forty years apart is enough time to miss entire chapters of somebody’s life.

What I do remember is Bub’s BBQ in Sunderland, where we started getting together about eight years ago, every fall: Tim and Judy, Diane and me. A picnic table, plates of barbecue, a few beers, old stories, a lot of laughs. You could see how much Tim and Judy loved each other just in how they sat together. Bub’s closed this year after almost fifty (!) years in business. That’s the kind of detail that tells you how much time had actually gone by without me noticing.

Time moves faster than the people living in it ever plan for. Forty years can pass between two friends who never had a falling out, never had one bad conversation, and just stopped calling. If there’s someone you’ve been meaning to call, calling closes the gap. Waiting only grows it.

Diane and I will be at the celebration of Judy’s life next month. After that, I’m hoping we get back out fishing. He’ll probably still catch more brookies than me. I don’t think I’ve got forty years left to find out.

Authentication, Not Encryption, Is Now The Big Quantum Worry

In the chapter on the threat to encryption in Quantum from the Ground Up, I cited a 2025 estimate from Craig Gidney at Google Quantum AI that RSA-2048 could be factored by a quantum computer with roughly one million noisy physical qubits, and noted that most experts placed Q-Day, the point at which a quantum computer can break current encryption, at around 2035. That estimate did not survive the spring.

In late March and early April 2026, two independent research results arrived within weeks of each other. Google published a major improvement to the quantum algorithm used to break elliptic curve cryptography, the math behind most of the internet's key exchange. Then Oratomic, a quantum computing startup, published a resource estimate suggesting RSA-2048 and P-256 could be broken with as few as 10,000 qubits on a neutral atom machine. That is roughly two orders of magnitude below the prior million-qubit estimate.

Why 10,000 Qubits Instead of a Million

The efficiency gain comes from error correction overhead, not from a change in the underlying mathematics. Shor's algorithm, the quantum algorithm that breaks RSA and ECC, has not changed. What changed is how many physical qubits it takes to build one reliable logical qubit on a neutral atom platform. Reporting on the Oratomic estimate put the ratio at roughly three to four physical qubits per logical qubit, a dramatically smaller overhead than superconducting platforms have required.

That ratio matters because it is the number that determines whether an attack is a research curiosity or an engineering project with a budget and a timeline. A million-qubit machine is not something any organization is building in this decade. A 10,000-qubit machine is in the range that QuEra, Atom Computing, and Pasqal have all stated as roadmap targets for 2026 to 2028 in the neutral atom chapters of this book. The number did not just get smaller. It got small enough to be plausible on hardware roadmaps that already exist.

How the Industry Responded

Cloudflare's response is the clearest signal of how seriously the industry is taking this. Cloudflare secures a significant fraction of global internet traffic, and the company had already completed most of its post-quantum encryption rollout, protecting against harvest-now-decrypt-later attacks where adversaries collect encrypted traffic today and decrypt it once a quantum computer is available. That work was largely done. The Google and Oratomic results changed what Cloudflare is worried about next.

Cloudflare's senior product director told reporters that authentication, not data confidentiality, is now the bigger concern. The distinction is worth sitting with. If a quantum computer can forge access credentials, an attacker does not need to decrypt anything. They can log into systems they are not supposed to have access to, or compromise a software update channel directly. That is a more immediate and more damaging failure mode than data theft, and it requires a different and more complex migration than swapping out an encryption algorithm.

Cloudflare set 2029 as its target for full post-quantum security across its platform, including authentication, and has accelerated its existing roadmap to hit that date. Google made a similar 2029 commitment. Coverage of the shift described the timeline change as a direct response to the two breakthroughs, and one report quoted the view that a coordinated attack against a high-value target, what the article called a moonshot attack, could plausibly arrive by 2030.

What This Changes in the Book

Chapter 13 of the first edition gave a 2029 to 2033 range as the accelerated estimate, driven by IonQ's October 2025 fidelity result. That range still holds, but the reasoning behind it has shifted. The IonQ result was a hardware fidelity improvement on an existing approach. The Google and Oratomic results are an algorithmic and architectural efficiency gain that reduces the qubit count needed for the same outcome. Both push in the same direction, but they are different kinds of progress, and the second edition will need to explain both rather than treating the timeline as a single number that moves around.

The Oratomic estimate specifically relies on high-rate quantum error-correction codes to achieve its 3:1 physical-to-logical overhead ratio. It is worth emphasizing that we are talking about 10,000 physical qubits here. Because the cybersecurity industry has spent a decade hearing that breaking RSA requires millions of physical qubits, highlighting that this breakthrough brings the physical hardware requirements down to the low thousands underscores just how radical this architectural shift is. Furthermore, this timeline collapse wasn't driven by hardware breakthroughs alone; as noted in the landmark paper co-published by Caltech and Oratomic, the team heavily leveraged AI-assisted algorithmic design and reconfigurable arrays to optimize these error-correcting codes, proving that classical AI is actively accelerating the software running on quantum roadmaps. Meanwhile, Google Quantum AI's joint research demonstrated that the logical qubit threshold to break elliptic curve cryptography could be slashed by 20x, creating a compounding effect where hardware requirements dropped while algorithm efficiencies spiked.

The organizational guidance from Chapter 13 does not change, it gets more urgent. Inventory your cryptographic dependencies. Specify quantum-resistant encryption in procurement. Re-encrypt long-term sensitive data. Build crypto agility into your architecture so algorithms can be swapped without a system rebuild.

The new addition, following Cloudflare's formal roadmap acceleration announcement, is that authentication systems deserve the same attention as encrypted data. As industry infrastructure leaders begin locking down their platforms against these newly compressed timelines, a compromised root certificate or access token is recognized as a far more catastrophic failure than a decrypted file, and it is the exact vulnerability the industry is now scrambling to patch.

This post will be incorporated into the second edition of Quantum from the Ground Up, out September 1. The full book, updated quarterly, is free at gordostuff.com/p/quantum-from-ground-up-hardware.html

The Classical Engine Inside Every Quantum Computer

Before writing this, I realized NVIDIA Ising already showed up in Chapter 12 of the book, the one on AI-assisted calibration and error correction. That chapter only had NVIDIA's side of it. A few days after that post went up, IEEE Spectrum ran a piece by Edd Gent that talked to Q-CTRL, Riverlane, IBM, and Google about the same problem, and the picture got more interesting. This post is what Chapter 12 looks like with that reporting folded in.

I have spent the past several months writing about quantum computing. Topological qubits, neutral atoms, silicon spin, photonics. Each one has its own physics, its own failure modes, its own argument for why it will win. Putting the book together, I noticed something missing from every chapter: what runs the thing once you turn it on.

The answer is classical computers. A lot of them. A recent IEEE Spectrum piece by Edd Gent lays out why, and it changes how I think about the supply chain side of this industry too.

Qubits are not reliable the way transistors are. A transistor comes off the line and works, billions of times, with no babysitting. A qubit drifts. It decoheres. It needs constant tuning just to stay usable, and that tuning happens in two stages. The first, called “bring up,” determines the frequency each qubit resonates at, how long it holds its quantum state, how sensitive it is to control pulses, and how strongly it interacts with its neighbors. Every one of those numbers affects error rates and how the qubit responds to control signals.

Done by hand, bring up still requires someone with a PhD and can take days or weeks, according to Jay Guilmart, lead product manager at Q-CTRL. The process resists scripting because each step depends on the result of the one before it. So Q-CTRL built software that examines each measurement, diagnoses what failed, and decides whether to proceed, repeat, or back up, rather than running a fixed sequence. Calibration does not end once a machine is running either. Parameters drift over time, so the system needs “runtime recalibration” to nudge things back into spec. But every cycle spent recalibrating is a cycle not spent running a circuit. As Guilmart puts it, if recalibration eats your uptime, the high fidelity you maintained is worthless.

Then there is error correction. Quantum information gets spread across many physical qubits to form a “logical qubit,” so that errors in individual qubits can be caught and fixed without destroying the encoded information. Measuring a qubit directly collapses its state, so error detection works indirectly through “parity checks,” which compare pairs of qubits to see if they agree. The pattern of agreements and disagreements is called a syndrome, and classical algorithms called decoders read that syndrome to locate errors.

This has to happen fast. Superconducting and silicon spin qubits hold their state for only microseconds to milliseconds, so decoding has to finish inside that window or the algorithm stalls. That rules out general-purpose processors. Decoders run on FPGAs or ASICs, chips built specifically for speed, according to Jerry Chow, CTO of quantum-centric supercomputing at IBM.

This is where the AI argument gets interesting, and where the field splits. In April, Nvidia released two models aimed at this problem. One uses a vision-language model to read calibration plots and feed an AI agent that decides how to adjust the processor. The other uses a convolutional neural network to catch the simple, localized errors that make up most faults, passing only the harder cases to a traditional decoder, for roughly a 2x speedup. Sam Stanwyck, Nvidia's director of quantum product, makes the case for AI on inference speed: models take time to train, but once trained they run fast and parallelize across chips as qubit counts grow.

Marco Ghibaudi, VP of engineering at Riverlane, pushes back on a different axis. Running anything on a GPU adds latency, even with massive throughput. His framing: you can have a very fat pipe, but if it is also a very long pipe, the data still arrives late. Riverlane's approach has been to shorten the pipe itself and make every stage faster, rather than chase raw throughput.

IBM's Chow lands in a similar place on calibration. He thinks AI shows real promise for understanding new architectures or unfamiliar circuit types, where you do not yet know what you are looking for. But for a well-characterized device where you are hunting small deviations from a known baseline, simpler physics-informed methods are cheaper and faster. Google's Adam Zalcman frames the two approaches as complementary rather than competing: neural networks are good at finding hidden patterns in syndrome data that handwritten decoders miss, so Google is building architectures that can run both, including its AlphaQubit 2 model.

Andi Gu, a Harvard PhD student working on AI decoders, takes the long view. He expects the “bitter lesson” to apply here eventually, the same pattern that played out in other parts of AI: general learning methods, given enough data and a large enough model, outperform hand-built algorithms over time. The obstacle right now is latency. Gu's group is working on shrinking AI decoders enough to fit on an FPGA, though smaller models trade off some accuracy, and finding that balance is still unresolved.

What stands out reading this against the QSCA work is how much of the “quantum workforce” conversation undersells this layer. When people talk about quantum jobs, they tend to mean physicists and people who understand qubits. But calibration software, decoder design, FPGA and ASIC engineering, and the AI models being built to assist both are squarely classical computing jobs. Guilmart's own warning, that calibration overhead will “blow up” past a thousand qubits and that current techniques will not scale, is as much a hiring problem as a technical one. Nobody in the article claims to have this solved, and Guilmart says plainly that no one is winning this battle today.

Every qubit type I have written about, no matter how exotic, sits on top of a rack of classical silicon doing the unglamorous work of keeping it alive. 

*****

Download Quantum from the Ground Up, a free PDF now available at gordostuff.com. This first edition covers posts through June 2026 and runs 50 pages across 19 chapters. It is written for someone entering the quantum workforce, or seriously considering it, who has a technical background but has not taken a graduate course in quantum mechanics.

Quantum from the Ground Up: First Edition Now Available

I have been learning and writing about quantum computing on gordostuff.com since October 2025. The posts started as a way to make sense of announcements as they happened: IonQ hitting 99.99% two-qubit gate fidelity, Microsoft unveiling Majorana 2, IBM running a 12,635-atom protein simulation, Apple releasing a 50,000-step formal proof of its post-quantum cryptographic library. Each one went up as a standalone post. After eight months and seventeen posts, it made more sense to put them together in one place.

The result is Quantum from the Ground Up, a free PDF now available at gordostuff.com. This first edition covers posts through June 2026 and runs 50 pages across 19 chapters. It is written for someone entering the quantum workforce, or seriously considering it, who has a technical background but has not taken a graduate course in quantum mechanics.

Download: Download PDF

The book covers six qubit fabrication platforms in sequence: superconducting circuits, trapped ions, photonics, neutral atoms, silicon spin qubits, and topological qubits. Each chapter describes the physical mechanism, the fabrication or trapping method, current performance numbers, and the specific engineering problems that remain unsolved. The qubit chapters are followed by sections on hybrid classical-quantum protein simulation, AI-assisted hardware calibration, post-quantum cryptography, and the hardware landscape as it stands in mid-2026.

Two chapters address the workforce directly. The first maps the adjacent skill sets the quantum supply chain actually needs: semiconductor process engineering, cryogenic systems operation, RF electronics, machine learning for hardware calibration, and cryptographic implementation. None of those require a physics PhD. The second chapter maps the full degree pathway from a two-year associate degree through a PhD, with specific programs, what each credential prepares you to do, and salary ranges at each level. The field has a real workforce shortage. The shortage is not only at the PhD tier.

The plan is to update the book quarterly as new posts are published. The field moves fast enough that a quarterly cycle makes sense: slow enough to let any given announcement settle, fast enough to stay current with actual engineering progress. This edition covers October 2025 through June 2026. The next update will incorporate posts from Q3 2026.

Seventeen posts, fifty pages, fifteen images, one dark navy cover. The posts were worth writing individually. They are more useful together. Download free at Download PDF.

Support This Work

The book is free and will stay free. If you find it useful and want to support future editions, you can contribute at ko-fi.com/gordostuff.

Topological Qubits: A Different Way to Build a Quantum Computer

Earlier posts here covered five qubit platforms: superconducting, trapped ion, photonic, neutral atom, and silicon spin. Each one answers the same question differently: how do you isolate a quantum system long enough to do useful computation? Topological qubits are the sixth approach, and the most contested. They exist at the intersection of materials science, condensed matter physics, and a twenty-year bet by Microsoft that the rest of the field is solving the wrong problem.

Every qubit in a quantum computer is fragile. Superconducting qubits, the kind Google and IBM use, operate near absolute zero and still lose coherence in microseconds. Gate error rates run between 0.1% and 1%, which sounds small until you consider that a useful fault-tolerant quantum computer may need error rates below one in a million. The industry has spent years stacking error correction on top of error correction, adding physical qubits to protect logical ones.

Topological qubits take a different approach. Instead of fighting noise with redundancy, they aim to make the qubit itself resistant to local disturbances. The physics relies on Majorana zero modes, exotic quasiparticles that store quantum information non-locally across two spatially separated points. Because the information is spread out, a local disturbance at one point cannot corrupt the qubit on its own. Topology, the branch of math that describes properties preserved under continuous deformation, gives the qubit its protection. You would have to disturb both ends of the system simultaneously to flip the state, which is far less likely than a single-point noise event.

Microsoft has pursued this path for nearly two decades through its Station Q research group. In February 2025, the company announced Majorana 1, described as the world's first quantum processor built on a topological core architecture. The chip used indium arsenide and aluminum, a hybrid superconductor-semiconductor platform. The research appeared in Nature in February 2025. The announcement drew immediate scrutiny. Independent researchers questioned the Topological Gap Protocol Microsoft uses to confirm the presence of Majorana modes, a University of St Andrews physicist published a challenge to its validity in March 2025, and Scientific American noted that Microsoft had previously retracted a high-profile Nature paper in 2021 after outside experts found the data could have come from material imperfections rather than topological qubits.

On June 2, 2026, Microsoft announced Majorana 2 at its Build 2026 conference. The new chip replaces aluminum with lead in the superconducting material stack and redesigns the semiconductor structure. Microsoft reports a mean qubit lifetime of 20 seconds, with some measurements exceeding one minute. That is a claimed 1,000-fold improvement over Majorana 1. Gate operations run at one microsecond, and the qubit footprint is 1/100th of a millimeter. The company now targets a commercially useful scalable quantum computer by 2029, moved up from a prior estimate of 2033. The materials iteration was accelerated with Microsoft Discovery, the company's agentic AI platform for scientific research.

The physics community's response has been consistent with the pattern from Majorana 1. Outside experts say the topological approach still lacks sufficient independent verification. The 20-second coherence time is striking if accurate, since superconducting qubits typically decohere in around 100 microseconds. But coherence time alone does not confirm the topological mechanism Microsoft claims is responsible for it. The company has a history of bold announcements followed by retraction or significant revision, and that history shapes how the community reads each new result.

Compared to the five platforms covered in earlier posts, topological qubits occupy a unique position. Superconducting qubits and silicon spin qubits are fabricated systems with well-characterized error mechanisms. Trapped ions and neutral atoms offer long coherence times but slow gates. Photonic qubits avoid decoherence but struggle with deterministic interactions. Topological qubits, if the physics holds, would offer built-in error protection that reduces the overhead for fault tolerance substantially. IBM's Condor chip reached 1,121 superconducting qubits in 2023. Microsoft is betting that the right number is not more physical qubits but better ones, and that topological protection is how you get there.

Whether Majorana 2 represents genuine progress or another contested milestone, Microsoft is the only major commercial player publishing peer-reviewed claims on topological qubit architecture. The debate in the physics community is real, the skepticism is well-founded, and the potential, if the approach works, remains significant. If you want the background on the other five platforms before going deeper on this one, the earlier posts are linked at the top of this page.