Grindr CEO George Arison reported that AI tools increased the company’s engineering output by 2.5 times between July 2025 and April 2026, with minimal growth in its engineering staff. In a shareholder letter, Arison estimated that achieving this output increase without AI would have required approximately 200 additional engineers and $60 million in annual costs.
During a second-quarter earnings call, Arison later stated the actual multiplier was 3.5 times, but Grindr adjusted the figure to 2.5 times, calling the higher estimate “unreasonable.” The company measures output by the volume of code shipped, a metric critics argue prioritizes quantity over quality.
Grindr’s AI spending is projected to reach $6 million on tokens in 2026, according to Arison. When asked about costs, he emphasized that the company focuses on return on investment (ROI) rather than token expenses. “We are of the view that people should use all the tools that are out there and not really worry about the cost to them as long as the ROI that we want to see is there,” Arison said.
The company’s AI adoption allows it to reallocate its existing engineering talent toward areas requiring human creativity and judgment, Arison noted. He added that the tech industry’s primary constraint remains a scarcity of exceptional engineering talent, making AI a strategic tool to maximize productivity without proportional hiring.
Critics question the validity of using code volume as a productivity metric. Marco Argenti, Goldman Sachs’ chief information officer, previously stated that measuring output by lines of code is “not really a great way to do it.” Alternative approaches, such as tracking AI token usage, have also faced scrutiny for potentially encouraging inefficiencies like “tokenmaxxing,” where developers prioritize token expenditure over meaningful output.
Grindr’s experience reflects broader industry trends, where companies explore AI’s role in software development. While some executives highlight AI’s potential to reduce labor costs and accelerate projects, others caution against over-reliance on metrics that may not fully capture software quality or long-term sustainability.