Google forms strike team as AI coding tools lag behind Anthropic's Claude Code dominance

Google Scrambles to Close AI Coding Gap with Anthropic

Google DeepMind has formed a strike team of researchers and engineers to develop AI coding models that can compete with Anthropic's increasingly dominant Claude Code platform 1

1

4

4

. The specialized team, led by research engineer Sebastian Borgeaud, aims to build AI coding models from scratch with capabilities spanning complex coding work, long-horizon programming, and writing entire software applications autonomously 4

4

5

5

. Google co-founder Sergey Brin and DeepMind CTO Koray Kavukcuoglu are directly involved in the initiative, underscoring the urgency within Google to catch up with competitors in the lucrative enterprise market for AI coding tools 4

4

5

5

.

Internal Concerns Mount as Engineers Prefer Rival Tools

Despite Google's public confidence, internal anxiety is mounting across the company, particularly within Google DeepMind, where some engineers prefer using Anthropic's Claude Code over Google's own solutions 1

1

. The company's Gemini AI models capabilities are scattered across half a dozen different coding products with inconsistent branding, reflecting a lack of focus and competing internal efforts that have hampered progress 1

1

. While most employees are banned from using competing tools due to security concerns, some teams working on Gemini, internal applications, and open source models have secured exceptions to use Claude Code 1

1

. This internal AI adoption challenge highlights a broader problem: even as businesses realize that AI coding tools can enable anyone to build products by prompting a chatbot, Google doesn't have a clear solution for them 1

1

.

AI Generates New Code at Accelerating Rate

Google announced that AI generates 75% of its new code as of April 2026, a sharp increase from 50% in late 2025 and just 25% in October 2024 2

2

. CEO Sundar Pichai attributed this surge to the deployment of agentic workflows, where engineers orchestrate systems that carry out complex tasks with minimal intervention 2

2

. The company cited productivity gains, noting that a particularly complex code migration completed by agents and engineers working together finished six times faster than was possible a year ago with engineers alone 2

2

. These systems now handle codebase migrations, large-scale refactoring, and other operations that historically required sustained human coordination 2

2

.

Why AI Coding Matters for Future Model Capabilities

Industry experts view AI coding as more than just a profitable early application—it's considered the key to building software that matches human capabilities 1

1

. "From a computer science point of view, if you win at coding this year, you get the raw data you need to win at model capability next year," said Raj Gajwani, former Google executive and current chief business officer at startup OpenArt AI 1

1

. Keith Zhai, co-founder of startup TinyFish, noted that "coding is the single easiest way to actually make money," adding that many Silicon Valley engineers toggle between Claude Code and OpenAI's Codex to see which program delivers better results, but Google often isn't part of that conversation 1

1

.

Consolidation Efforts and Strategic Repositioning

To address internal confusion over priorities, Chief AI Architect Koray Kavukcuoglu is working with Google's main engineering team to unite the company's internal AI coding tools under Antigravity, a platform released last year 1

1

. Google DeepMind is also devoting more resources to AI coding, with Nobel Prize winner John Jumper reportedly working on code generation alongside the new team 1

1

. In an internal memo, Sergey Brin emphasized the need to "urgently bridge the gap in agentic execution and turn our models into primary developers," asking engineers working on Gemini to use internal agents for complex tasks 4

4

. The company is even tracking employee usage of AI coding tools through an internal leaderboard, in some cases tying their use to performance reviews 2

2

5

5

. While Google still has deep pockets, substantial computing power, and has made significant strides in foundation model quality, Silicon Valley engineers are embracing AI coding so quickly that even a momentary lag could prove consequential in this rapidly evolving software development landscape 1

1

.