Sundar Pichai Defends Google AI Strategy Amid Gemini 3.5 Pro Delay

Alphabet CEO Sundar Pichai mounted a forceful defense of Google's artificial intelligence strategy during the company's July 22 earnings call, pushing back against mounting investor concerns that the tech giant has fallen behind rivals following the delay of its flagship Gemini 3.5 Pro model. The delay of Gemini 3.5 Pro, originally slated for a June release following its announcement at Google I/O in May, has become a flash point for anxiety about Google's competitive position in the rapidly evolving AI landscape. The model, expected to bolster Google's standing in AI coding and autonomous "agent" tasks—two of the industry's most fiercely contested areas—remains in partner testing with no public release date announced. 

Internal Frustration and Technical Hurdles- According to Bloomberg reports citing ten current and former employees, the delay has sparked significant frustration among Google engineers, AI researchers, and management. The core issue appears to center on the model's programming capabilities. Google updated Gemini's training data in late June specifically to improve coding performance, but the results have reportedly fallen short of expectations. The internal challenges run deeper than technical shortcomings. Sources describe a complex organizational landscape where multiple stakeholder approvals are required before model releases, and AI capabilities must be integrated across Google's vast product ecosystem including Search, Maps, and YouTube. One former employee characterized aligning various business units as akin to "boiling the entire ocean Competing pressures have created internal friction. Google Cloud, DeepMind, and the Android team are all developing AI coding tools for developers, with some consumer product teams also participating. This redundancy has made it difficult for individual projects to secure adequate resources and attention.

Pichai's Strategic Defense- Rather than dwell on the Gemini 3.5 Pro delay, Pichai used the earnings call to articulate a broader vision for Google's AI ambitions. He acknowledged areas requiring improvement—specifically naming coding and "agent coding" as priorities—while emphasizing that Google remains at the frontier across many AI attributes. Pichai repeatedly highlighted the Gemini Flash family as Google's practical "workhorse" models, emphasizing their cost-effectiveness and speed. He pointed to Gemini 3.6 Flash, released the same week, which improved by more than 10 points on a coding benchmark compared to its predecessor while using fewer tokens. The company also unveiled Gemini 3.5 Flash-Lite and a cybersecurity-focused Flash Cyber model, demonstrating a strategy of deploying specialized models across multiple use cases including customer service, data analytics, and enterprise software. Looking ahead, Pichai described Gemini 4 as a "very ambitious effort," stressing that Google is training a significantly larger model designed to compete at the frontier. He disclosed that Gemini 4's roadmap includes rolling out models "almost at a monthly cadence," suggesting an accelerated release schedule once the next-generation foundation is established.

Investor Skepticism- Despite Pichai's assurances, Wall Street has responded with caution. Alphabet's shares fell more than 3% in after-hours trading following the earnings call and remain down approximately 9% since late April. This decline reflects concerns over Gemini's delays compounded by several high-profile executive departures. The investor skepticism comes even as Google Cloud posted another quarter of blistering growth—up 82%, far exceeding the average estimate of 64%. However, escalating capital costs have raised additional concerns. Alphabet raised its capital plans by $15 billion to a range of $195 billion to $205 billion.

The Organizational Challenge- The delay has exposed deeper organizational challenges within Google's AI operations. Sources describe a company where shifting strategies and competing internal factions have complicated unified direction. Google has taken steps to address this fragmentation, with Chief AI Architect Corey Kavukcuoglu working with major engineering teams to unify internal AI coding tools, and DeepMind establishing a dedicated AI coding team earlier this year. Cultural factors have also played a role. Some Google engineers have historically maintained conservative views that all important code should be human-written to meet Google's exacting standards. Early restrictions on using Gemini for coding tasks—implemented over concerns about proprietary code leaking into training data—limited engineers' exposure to AI-assisted development. While those restrictions have since been relaxed, and Google now reports that approximately 75% of new code is AI-generated with human review, the early caution may have slowed internal adoption.

Looking Ahead- Pichai's robust defense underscores a growing challenge facing Google: persuading investors to judge its AI ambitions not by the delay of a single flagship model, but by the scale of an ecosystem spanning cloud infrastructure, custom AI chips, and a portfolio of Gemini models embedded across its products and services. The company has also emphasized its engagement with the U.S. government on model testing and broader regulatory frameworks, suggesting a commitment to responsible development amid increasing scrutiny of frontier AI systems. Whether Pichai's strategic reframing can reassure investors remains to be seen. As one former employee noted, some researchers have already departed for Anthropic and other leading AI labs, disappointed with Google's position in the AI race. For now, Google's AI strategy rests on a bet that ecosystem breadth and long-term technical investment will ultimately prevail over the competitive metrics of any single model release.