“Tokenmaxxing” is making developers less productive than they think

There’s an aged saw successful management: What you measurement matters. And, typically, you get much of immoderate you’re measuring.

Software engineers person debated productivity metrics for decades, starting with lines of code. But arsenic the caller procreation of AI coding agents delivers much codification than ever, what their managers ought to beryllium measuring is little clear.

Enormous token budgets—essentially, the magnitude of AI processing powerfulness a developer is authorized to consume—have go a badge of grant among Silicon Valley developers, but that’s a precise weird mode to deliberation astir productivity. Measuring an input to the process makes small consciousness erstwhile you presumably attraction much astir the output. It mightiness marque consciousness if you’re trying to promote much AI adoption (or selling tokens), but not if you’re trying to go much efficient.

Consider the grounds from a caller people of companies operating successful the “developer productivity insight” space. They’re uncovering that developers utilizing tools similar Claude Code, Cursor, and Codex make a batch much accepted codification than they did before. But they besides find that engineers person to instrumentality to revise that accepted codification acold much often than before, undercutting claims of accrued productivity.

Alex Circei, the CEO and laminitis of Waydev, is gathering an quality furniture to way these dynamics; his steadfast works with 50 antithetic customers that employment much than 10,000 bundle engineers. (Circei has contributed to TechCrunch successful the past, but this newsman had ne'er met him before.)

He says that engineering managers are seeing codification acceptance rates of 80% to 90%—meaning the stock of AI-generated codification that developers o.k. and keep—but they’re missing the churn that happens erstwhile engineers person to revise that codification successful the pursuing weeks, which drives the real-world acceptance complaint down betwixt 10% and 30% of generated code.

The emergence of AI coding tools led Waydev, founded successful 2017 to supply developer analytics, to wholly rework its level successful the past six months to code the proliferation of accelerated coding tools. Now, the institution is releasing caller tools that way the metadata generated by AI agents, offering analytics connected the prime and outgo of their codification to supply engineering managers with much penetration into some AI adoption and efficacy.

While analytics companies person an inducement to item the problems they find, the grounds is mounting that ample organizations are inactive figuring retired however to usage AI tools efficiently. Major companies are noticing—Atlassian acquired DX, different engineering quality startup, for $1 cardinal past year, to assistance its customers recognize the instrumentality connected concern connected coding agents.

The information from crossed the manufacture tells a accordant story: More codification is being written, but a disproportionate magnitude of it isn’t sticking.

GitClear, different institution successful this space, published a report successful January that recovered AI tools accrued productivity, but besides that its information showed “regular AI users averaged 9.4x higher codification churn than their non-AI counterparts”—more than treble the productivity gains the tools provided.

Faros AI, an engineering analytics platform, drew connected 2 years of lawsuit information for its March 2026 report. The finding: codification churn—lines of codification deleted versus lines added—had accrued 861% nether precocious AI adoption.

Jellyfish, which bills itself arsenic an quality level for AI-integrated engineering, collected data connected 7,548 engineers successful the archetypal 4th of 2026. The steadfast recovered that the engineers with the largest token budgets produced the astir propulsion requests (proposed changes to a shared codebase), but the productivity betterment didn’t scale. They achieved 2 times the throughput astatine 10 times the outgo of tokens. In different words, the tools are generating volume, not value.

These kinds of statistic ringing existent erstwhile you speech to developers, who are uncovering that codification reappraisal and method indebtedness are stacking upm adjacent arsenic they revel successful the state of the caller tools. One communal uncovering is the quality betwixt elder and inferior engineers, with the second accepting acold much AI-generated code, and dealing with a larger magnitude of rewriting arsenic a consequence.

Still, adjacent arsenic developers enactment to recognize precisely what their agents are up to, they don’t expect turning backmost anytime soon.

“This is simply a caller epoch of bundle development, and you person to adapt, and you are forced to accommodate arsenic a company,” Circei told TechCrunch. “It’s not similar it volition beryllium a rhythm that volition pass.”