Alphabet fell 5%, its worst day in a year, after two top researchers quit for OpenAI and Anthropic

• Noam Shazeer, vice president of engineering and co-lead of Gemini, quit to join OpenAI after returning to Google less than two years ago.
• John Jumper, Nobel Prize winner and senior research scientist at DeepMind, departed for Anthropic on Friday, marking two high-profile exits in one week.
• Google plans to spend $180 billion to $190 billion in fiscal 2026 on data centers and AI computing, with capital spending rising further in 2027.

• 5% Alphabet stock decline on Monday versus worst day in over one year prior
• $180-190B Planned fiscal 2026 capital expenditure for data centers and AI infrastructure
• 1% Google search traffic decline as ChatGPT traffic gains in recent weeks

Alphabet closed down 4.9% on Monday, marking the steepest single-day loss since a roughly 7% decline in May 2025, and outpacing broader tech weakness across the Nasdaq and other major peers.

The timing proved critical: the stock slide came just days after Noam Shazeer announced his departure to OpenAI on Wednesday, followed by John Jumper’s Friday announcement that he would leave DeepMind for Anthropic.

Both moves underscore a fundamental shift in the competitive dynamics of artificial intelligence development, one where Google, despite its historical advantages in computing infrastructure and talent, is now losing marquee researchers to better-capitalized rivals.

Google’s Gemini Co-Lead Defects to OpenAI After Brief Return

Shazeer’s exit carries particular sting because it represents a second departure. He had rejoined Google less than two years ago, suggesting the company had successfully lured him back from industry opportunities. As vice president of engineering and co-lead of the Gemini AI models, Google’s flagship response to ChatGPT, his role positioned him at the center of the company’s AI strategy.

His jump to OpenAI, the entity behind the ChatGPT phenomenon that has reshaped investor sentiment toward generative AI, signals that Google’s internal environment or compensation cannot match the pull of leading-edge AI labs elsewhere.

The departure carries symbolic weight beyond a single hire or resignation. Shazeer’s decision to leave again after a recent return suggests that high-level AI researchers now view the competitive landscape as driven by access to cutting-edge models, computational resources, and the prestige of working at firms seen as advancing the frontier of AI capability.

OpenAI’s valuation and funding position, bolstered by backing from Microsoft and other major investors, may have offered both financial and technical advantages that outweighed Alphabet’s scale.

Jumper’s Nobel Prize credential adds another dimension to the talent drain.

Nobel Laureate Jumper Leaves DeepMind for Anthropic

John Jumper, who won the Nobel Prize in Chemistry for computational protein structure prediction, occupied one of the most prestigious research roles at Google DeepMind. His departure to Anthropic on Friday capped a bruising 48-hour period for Alphabet’s institutional narrative.

Jumper’s stature as a world-class scientist, rather than an engineering manager, indicates that even researchers at the apex of the AI field believe better work is happening, or will happen, outside Google’s walls.

Anthropic, the AI safety-focused competitor founded by former OpenAI researchers including Dario and Daniela Amodei, has positioned itself as an alternative pole of gravity for talent. While smaller and younger than Google, Anthropic has attracted substantial venture and strategic investment precisely because it represents a perceived vanguard in large language model development.

Jumper’s move suggests that prestige and frontier-level work, not size or legacy, now dominate the calculus for elite researchers.

These two defections arrive amid a broader pattern of attrition at Google’s AI units.

Talent Exodus Exposes Google’s Vulnerability in the AI Arms Race

Market analysts have noted that the two exits represent not isolated incidents but a symptom of structural disadvantage. Google’s own AI teams have experienced a series of layoffs in recent months, which likely diminished morale and signaled uncertainty about the company’s strategic direction or resource allocation within AI research.

When layoffs and departures occur simultaneously, they compound each other: remaining staff worry about future stability, and external talent becomes more hesitant to join a team perceived as declining.

The competitive context sharpens the problem. Last year, Google briefly held the perception of leading the AI race with innovative methodologies, but well-capitalized competitors and emerging open-source models have rapidly closed gaps.

Traders on Polymarket currently assign only 13% odds that a Google model will rank as the best artificial intelligence by the end of July, a metric that, while imperfect, reflects market expectations about relative capability. The low probability underscores investor sentiment that Google no longer occupies the technological frontier it once dominated.

Retaining researchers becomes impossible when the broader market doubts a company’s technological leadership. Shazeer and Jumper likely perceived that OpenAI and Anthropic, not Google, represent the most intellectually fertile and strategically important environments for AI research in 2025.

ChatGPT Erosion in Google Search and Rising Competitive Threats

Beyond talent, the departures highlight existential threats to Google’s core business. Since OpenAI’s launch of ChatGPT, which surpassed one billion monthly active users recently, generative AI has posed a direct challenge to Google’s search dominance.

The risk operates on two fronts: first, users may shift away from Google Search to AI chatbots for information retrieval; second, Google’s attempts to defend search with AI features risk cannibalizing its own advertising business, which depends on users clicking through search results.

Early data signals cracks. Google’s search traffic has declined by more than 1% over recent weeks, while ChatGPT traffic has grown. DuckDuckGo, a privacy-focused search engine, is promoting a “no-AI” positioning through new browser features and experiencing measurable install growth as a result.

Meanwhile, open-source Chinese models such as DeepSeek and z.AI are gaining traction because they offer near-parity capability to leading US models at substantially lower cost, a dynamic that could compress margins across the industry and reduce the economic justification for Alphabet’s massive capital expenditures.

For investors, the search erosion is not theoretical.

Google’s Ballooning AI Capex Faces Margin Pressure and Unproven ROI

Alphabet has announced plans to spend $180 billion to $190 billion in fiscal 2026 alone, with the majority directed to data centers and AI computing infrastructure. More concerning for equity investors, the company has cautioned that capital spending will increase further in 2027.

This capital intensity arrives at a time when search revenue, Alphabet’s primary profit engine, faces headwinds from both ChatGPT adoption and lower-cost open-source alternatives.

The financial math now dominates institutional discussions. Investors are beginning to question whether the AI spending boom can deliver returns commensurate with deployment. If Google’s search business faces structural decline, then the billions invested in AI infrastructure may not generate new revenue streams fast enough to offset margin compression.

The stock market’s reaction, a 5% decline, significantly worse than broader tech, reflects skepticism about whether Google can monetize its AI capabilities before its core business erodes.

Capital expenditure at this scale also constrains flexibility. Google’s operating cash flow, while substantial, will be stretched by $180 billion-plus annual capex commitments. Shareholders face a choice: either the company generates breakthrough AI products that justify the spend, or it has made a massive bet that fails to pay off within a reasonable timeframe. The departures of Shaz

Original reporting: cryptopolitan.com