Alphabet CEO refutes claims that Google is losing ground in AI race
Alphabet CEO Sundar Pichai strongly defended the company during an earnings call on Wednesday ’s artificial intelligence (AI) strategy, countering concerns that the company is falling behind due to delayed flagship model releases and being overtaken by rivals in AI programming. Investors' unease is growing as Google delays the release of Gemini 3.5 Pro. The model, originally scheduled to be launched in June, had high hopes and was intended to boost Google's competitiveness in AI programming and autonomous "agent" tasks, two of the most competitive areas in the industry. The latest delay further stoked market concerns: Google appears to be losing momentum just as OpenAI, Anthropic and many Chinese competitors accelerate the pace of model releases. On Wednesday's conference call, Pichai took an uncharacteristically defensive stance when analysts asked about the status of Google's cutting-edge models and whether it could still remain competitive at the forefront of AI research and development. Analyst Doug Anmuth questioned whether Gemini can continue to maintain industry-leading competitiveness and Google's slow release of models. Pichai said: "We have obviously always had models that are at the cutting edge. In many ways, we are still at the forefront of the industry; there are also some areas where we admit that we need to improve, and programming capabilities and agent programming are one of them." Pichai didn't talk much about the delayed release of the Gemini 3.5 Pro, but repeatedly emphasized the Gemini Flash, Google's cheaper and faster "workhorse" model. The model has been widely used in ever-expanding application scenarios such as network security, customer service agents, data analytics, and enterprise software. He specifically mentioned Gemini 3.6 Flash, released this week, saying that its performance in programming benchmarks has improved by more than 10 percentage points compared with the previous version, while using fewer tokens. Google also released Gemini 3.5 Flash-Lite and a Flash Cyber model aimed at the cybersecurity field on Tuesday, while Gemini 3.5 Pro is still in partner testing. "I think everyone will be satisfied when Google releases Gemini 4," Pichai said, calling it "an ambitious effort." He emphasized that Google is training a much larger model with the goal of being able to compete on the cutting edge of technology once it is released. He added that the company remains "very determined and confident" in maintaining its leadership in AI. When Barclays analyst Ross Sandler raised similar concerns about Google's slow pace of model releases, Pichai revealed that Gemini 4's product roadmap plans to launch new models "almost once a month." Pichai's strong defense highlights a growing challenge facing Google: persuading investors to judge its AI ambitions not just by the delays of a single flagship model, but by looking at its vast AI ecosystem - which spans cloud infrastructure, custom AI chips, and the matrix of Gemini models embedded in Google's various products and services. Alphabet shares fell more than 3% in after-hours trading on U.S. stocks. Concerns about delays in the Gemini project, coupled with the departure of several senior executives, have caused its stock price to fall by about 9% since the end of April.