Google CEO Sundar Pichai has openly admitted that the tech giant is currently trailing behind its rivals when it comes to the highly competitive frontier of AI-powered coding agents. In an interview published in The New York Times, Pichai addressed whether Google is lagging behind the leading edge in automated programming.
While defending Google’s broader AI capabilities, he acknowledged that competitors like Anthropic have captured a lead in helping developers manage complex, long-running software tasks. However, he emphasized that coding is fundamentally foundational to everything Google does. He expressed strong confidence that internal usage data and new model iterations will quickly close the gap.
“Look, coding ends up being very foundational in everything we do. I think it’s an important frontier to be on,” Pichai said. He added, “There are areas in coding where we have been very good. But in terms of these long-running tasks where serious developers are working on complicated code bases, we are making progress. But there is a gap to the frontier where others are. We are well aware of it and making progress there.”
Google models are ‘best’ for general intelligence
Pichai emphasized that while Google's models remain at the absolute frontier for general intelligence features—including text processing, multimodality, voice, audio, and overall reasoning—the company has hit a temporary bottleneck in specialized programming tools.
“When it comes to agentic coding with tool use, instruction following and long-horizon tasks, I think we are a bit behind at this moment,” Pichai confessed. He added, “But we are hard at work, and the space is so dynamic. All of the leading labs have their own pretraining cycles. You have these cadences, and they may not exactly match up.”
Sundar Pichai on what’s missing to be the leader
According to Pichai, part of the reason Google fell behind in agentic coding was a lack of direct user data. Competitors like Anthropic managed to capture the developer market early through highly popular interfaces.
“Coding was an area where getting access to the data flows was important. We maybe didn’t quite have the surface, like Claude Code as an example, or what Anthropic maybe had with Cursor, too,” Pichai explained.
To address this, Google has rolled out Gemini 3.5 Flash, which Pichai describes as a big step forward that directly addresses the areas where the company was lagging. “I think we took a big step forward with 3.5 Flash. It addresses some of the areas we have been behind in, and obviously getting it out in the real world and iterating with that data coming back is going to really help us,” Pichai noted.
Internal tools show promise
Despite being behind publicly, Pichai revealed that Google's internal development tells a vastly different story. Google engineers have been heavily dogfooding their own advanced coding AI, known internally as Antigravity 2.0.
“And so getting Antigravity 2.0, we have been using it internally at Google for a while. I shared the token usage at Google I/O. I have never seen anything like it internally; we are doubling every week, and people are really putting the models to work. That is helping us hill climb quite a bit. The frontier is very dynamic, but I am very, very optimistic and confident we will push through there,” Pichai said.
Ultimately, Pichai dismissed the idea that Google is out of the race, stating that while the company is behind in agentic coding at the moment, it will soon be at the forefront.
