Armstrong Pushes AI to Half of Coinbase Code

Coinbase, a publicly traded exchange since 2021, serves over 108 million verified users and lists more than 250 digital assets. Armstrong’s directive, issued earlier this year, mandated all engineers to adopt AI coding tools like GitHub Copilot and Cursor. Non-compliance led to individual reviews, with some engineers facing termination. The company secured enterprise licenses to ensure widespread access, emphasizing responsible use through mandatory human oversight, particularly for security-sensitive systems.

This strategy builds on Coinbase’s 2022 cost-cutting during a crypto market downturn, when it laid off 18 percent of its staff amid falling prices, now prioritizing technological edge over further layoffs. The firm’s aggressive adoption contrasts with competitors like Binance and Kraken, which primarily use AI for market analytics and fraud detection rather than direct code generation.

The company employs 4,279 people as of June 2025, mainly for technical roles. About 47 percent of job postings require AI skills. Engineers report slashing development timelines, with tasks like codebase refactoring now taking days rather than months.

Monthly training sessions extend AI use to design, finance, and decision-making teams, fostering an “AI-Native” culture where tools augment human efforts. According to Stack Overflow’s 2025 Developer Survey, 84 percent of developers are using or planning to use AI coding tools, with 51 percent actively incorporating them into workflows, up from 44 percent the previous year.

However, trust in AI outputs has declined, with only 33 percent of developers fully trusting accuracy, down from 43 percent in 2024. Coinbase’s 40 percent AI-generated code surpasses most peers in the crypto space, though tech giants like Microsoft report similar integration levels with tools like Copilot.

Coinbase has posted swift progress as AI accounted for just 15 percent of daily code in April 2025, but rose to 33 percent by August before hitting the current 40 percent mark. Armstrong has shared that this acceleration stems from mandatory adoption, which he described as “heavy-handed” but necessary to signal AI’s critical role in the company’s future.

Reports from consulting firm McKinsey indicate AI can boost developer productivity by up to 50 percent without leading to widespread job reductions, as workers pivot to higher-level strategic roles. In Coinbase’s case, this has enabled faster upgrades to systems handling user transactions and compliance, crucial in a sector where speed can mean competitive advantage.

Security risks remain a concern, but Coinbase stresses that AI is barred from critical financial infrastructure and subject to rigorous human review. The 2022 Ronin Network hack, which exploited code vulnerabilities and resulted in a $625 million loss from the blockchain gaming platform’s bridge, serves as a stark reminder of the stakes for crypto platforms. Competitors like Kraken and Gemini take a more cautious approach, using AI mainly for market analytics and security monitoring rather than code generation.

The crypto sector remains under constant regulatory pressure from agencies such as the SEC, while volatility continues to unsettle markets. Coinbase’s rapid timeline invites questions about whether speed is undermining safety, particularly as AI is scaled to manage rising transaction volumes, which reached $312 billion per quarter in early 2025. After Armstrong’s announcement, Coinbase shares dipped, reflecting investor caution over the risks of accelerating AI adoption in a high-stakes environment.

Armstrong’s 50 percent target has the potential to redefine AI’s role in finance, though execution relies on safeguards against flaws like those uncovered in Cursor. Security researchers reported that the tool could replicate malicious instructions from files such as README and LICENSE into active code, raising concerns about supply-chain attacks. The controversy has added weight to the ongoing debate over AI adoption, with Coinbase pushing further than most in a sector already defined by disruption.

Coinbase’s approach follows the sector’s pattern of testing limits. Today the focus is on AI, which management sees as a tool to surpass competitors. Security concerns remain, and the October deadline will reveal whether the balance between innovation and safety can be maintained. The shift also reflects a wider trend in hiring, with AI skills now appearing in nearly half of all industry job ads. Traditional banks like JPMorgan Chase and Goldman Sachs already use AI to monitor fraud and execute trading strategies across millions of transactions.

Coinbase’s plan to rely on AI for direct code generation sets it apart, making the outcome a test case for whether the benefits outweigh the risks of data leaks and exploit chains.