Section 174 and the vaporization of the tech labor market

The prevailing narrative across Silicon Valley holds that the sudden, violent collapse of the software labor market is the result of a toxic cocktail. Ask any venture capitalist or displaced developer, and they will point to OpenAI’s aggressive product roadmap, the definitive end of zero-interest-rate policy, and a coordinated "vibecession" engineered by founders eager to claw back leverage from a historically pampered workforce. I generally prefer to remain an observer in these cultural skirmishes, but the timeline on this consensus theory is hopelessly flawed. The devastation in tech hiring does not correlate to the moment we collectively realized large language models could generate boilerplate React components; it maps precisely to the expiration of a legislative grace period buried deeply inside a 2017 federal tax bill. We are witnessing the delayed, systemic detonation of IRS Section 174.

For well over a decade, Hacker News maintained the tech industry’s most reliable, remarkably un-gameable macroeconomic indicator: the ratio of the monthly "Who is Hiring" thread to its "Who Wants to Be Hired" counterpart. When you look at the Hacker News thread serving as the source dossier for current applicant despair, you are witnessing the acute macro failure of a heavily regulated system reacting rationally to new constraints. Independent analysis confirms that the volume of open roles in these threads dropped by over 50% since the peak in early 2022. What physical, structural shift occurred in the early months of 2022? The United States federal government fundamentally rewired the baseline cost of human labor in software engineering.

Historically, software developer salaries were treated by the tax code as immediately deductible research and development operational expenses. You spent a dollar on an engineer to build a web application, you deducted a dollar from your taxable revenue. Past tense. During the frantic drafting of the Tax Cuts and Jobs Act of 2017, legislators found themselves needing the math to score favorably in a Congressional Budget Office spreadsheet. To offset the revenue lost from corporate tax rate cuts, they added a delayed provision requiring companies to capitalize and amortize domestic SRE expenditures over five years starting in tax year 2022.

The working assumption inside the Beltway was that a future Congress would simply repeal this budgetary sleight of hand before it actually took effect. (They did not, which is a fairly predictable outcome for anyone familiar with the stochastic operational velocity of the legislative branch.) Now, formal administrative reality dictates that these expenses must be capitalized and amortized.

To fully internalize why this seemingly minor accounting technicality vaporized the job market, we have to follow the money. We will trace a single senior software engineering salary from the payroll processor into the corporate ledger, running it straight through the Section 174 meatgrinder via a "Phantom Profit" cash-flow waterfall simulation.

Imagine a reasonably successful, bootstrapped SaaS company. They earn exactly $150,000 in revenue this year and, being good capitalists intent on growth, they decide to reinvest all of it by hiring one senior backend engineer for $150,000. Under the old regime, their net profit for the year is zero. Consequently, their tax liability is zero. The system functions smoothly.

Now, consider that exact same company operating under the new Section 174 mandate.

The Phantom Profit Waterfall

• Gross Revenue: $150,000
• Actual Payroll Expended: -$150,000
• Operating Bank Balance: $0
• Tax Year 2021 (The Old Regime): $150k immediate deduction -> $0 Taxable Income -> $0 Tax Owed
• Tax Year 2022 (The Section 174 Regime): $15k allowed deduction -> $135k Taxable Income -> ~$28,000 Tax Owed

You might be wondering where that meager $15,000 deduction came from. The law includes a mandatory mid-year convention for the first year of amortization, operating on the polite bureaucratic fiction that all corporate R&D occurs exactly halfway through the calendar year. Consequently, companies are forced by statute to only deduct 10 percent of those costs in the first year.

Look closely at the resulting financial crater. Our hypothetical startup spent its entire bank account employing this engineer. But the Internal Revenue Service looks at this business, observes the capitalized asset that used to be considered a human being's labor, and declares that the company has $135,000 in taxable profit. At a roughly 21% blended corporate and state tax rate, the startup suddenly owes the federal government $28,350 in cold, hard cash. (Cash they decidedly do not have, because they already handed it to the engineer via direct deposit.) They are being heavily taxed on phantom profit.

This waterfall of phantom profit aggressively alters the incentive gradient of the entire technology ecosystem. When you read panicked reports about companies that are barely profitable, or even loss-making, suddenly have large tax bills, you are watching rational agents optimizing under overtly hostile tax constraints, rather than greedy executives engaging in a coordinated vibecession.

Put yourself in the shoes of a CFO at a mid-sized technology firm. Your job is to ensure the business does not default on its obligations to the state. If every new engineering hire generates a massive, unbudgeted cash tax liability that hits the balance sheet before the employee has even completed their onboarding ETL jobs, the mathematically correct survival mechanism is to hoard cash. Doing so requires freezing headcount, halting speculative infrastructure projects, and unceremoniously pulling approved job requisitions from your baffled engineering directors. You cannot afford the tax penalty of growth.

A skeptical founder might point out that the Section 41 R&D tax credit exists exactly for this reason, meant to cushion the blow for innovative firms. (Ask your accountant; founders frequently misunderstand the cascading rules here.) Under IRC Section 280C, if you claim the R&D credit, you must reduce your already-crippled capitalized deduction pool by the credit amount claimed. The friction remains; it does not save the math. You might also note that heavily venture-backed startups with zero revenue do not pay immediate cash taxes anyway. This is technically true, but they pay the iron price regardless by watching their vital Net Operating Loss (NOL) carryforwards evaporate before their eyes. This drastically alters their valuation math, terrifies their board, and forces down rounds. At every level of the capital stack, Section 174 makes the act of employing a software developer brutally, undeniably capital inefficient.

We have followed the math downward from a federal mandate, through the frantic adjustments of the corporate treasury, right down to the anxious applicants constantly refreshing Hacker News on a Tuesday morning. The state has fundamentally changed the definition of software development, moving it from a lightweight operational necessity to a severely penalized capital asset.

The market will adjust, because markets always do. But they will adjust to a structural reality where typing code into a terminal is treated, by the full force of the United States federal government, as identical to pouring concrete for a factory. We will see leaner engineering teams, a chilling effect on speculative features, and a lot of highly skilled people wondering why the system abandoned them. The bots didn't take the jobs; the accountants simply amortized them out of existence.