Overview

In a span of just three days in June 2026, Alphabet lost two of its most prominent artificial intelligence leaders: Noam Shazeer, co-lead of the Gemini model family, departed for OpenAI, and John Jumper, the Nobel Prize-winning co-creator of AlphaFold, announced his move to Anthropic. These exits are not isolated incidents but the latest and most dramatic manifestations of a sustained talent hemorrhage from Google’s AI ranks. Over the past several years, many of the researchers who built the foundational Transformer architecture—the very technology that underpins modern generative AI—have left Google to found or join competing labs. The cumulative effect is a growing asymmetry in top-tier AI talent between Alphabet and its chief rivals, OpenAI and Microsoft, with direct consequences for Google’s flagship AI products: Gemini, Google Cloud AI, and Search AI features.

This report examines the specific departures, their impact on Alphabet’s product roadmap, the company’s retention strategies, and how OpenAI and Microsoft are outmaneuvering Google in the war for AI talent. It also assesses the broader competitive landscape, including market share shifts, research output, and the spiraling cost of AI infrastructure, while noting counter-examples of resilience within Alphabet’s AI ecosystem.

The Talent Exodus: Key Departures and Their Impact

Noam Shazeer: Gemini Co-Lead Departs for OpenAI

Noam Shazeer, a Vice President of Engineering at Google and co-lead of the Gemini AI models, announced on June 18, 2026, that he was leaving the company to join OpenAI [2][14]. Shazeer is a co-author of the seminal 2017 paper “Attention Is All You Need,” which introduced the Transformer architecture and is widely credited with kickstarting the modern large language model era [2][7][14]. He originally joined Google in 2000 and spent 17 years there before leaving in 2021 to co-found Character.AI after Google declined to pursue a chatbot project he championed [2][5][14]. In August 2024, Google paid approximately $2.7 billion for non-exclusive rights to Character.AI’s technology, with an agreement that Shazeer would return to work at Google [2][5][7][14]. Less than two years later, he departed again.

At OpenAI, Shazeer will serve as lead for architecture research, focusing on core structural blueprints for future AI models [2][14]. OpenAI CEO Sam Altman publicly celebrated the hire, writing that Shazeer was “one of the people I have most wanted to work with since the very beginning of openai… only took 10 years. i think it will be worth the wait!” [11][14]. Google’s statement was notably restrained: “We are grateful for Noam’s meaningful contributions to Google over the years” [2][14]. Axios described the move as “a major win for OpenAI in the AI talent wars” [7][10].

The immediate impact on Gemini is significant. Shazeer was not merely a figurehead; he was the operational co-lead of Google’s most important AI product line. His departure creates a leadership vacuum at a moment when Gemini faces intense competition from OpenAI’s GPT-5.5, Anthropic’s Claude Opus 4.8, and Microsoft’s newly launched MAI model family. Although Gemini has grown to 900 million monthly users [27][33], the loss of its co-architect raises questions about the pace and direction of future development.

John Jumper: Nobel Laureate Leaves DeepMind for Anthropic

On June 19, 2026, just one day after Shazeer’s announcement, John Jumper—a Vice President and Engineering Fellow at Google DeepMind and co-winner of the 2024 Nobel Prize in Chemistry for AlphaFold—revealed he was joining Anthropic [3][4][6][15]. Jumper had spent nearly nine years at DeepMind, where he co-created AlphaFold, the breakthrough AI that has predicted over 200 million protein structures and transformed biological research [3][4][6]. DeepMind CEO Demis Hassabis, who shared the Nobel with Jumper, responded: “What we achieved with AlphaFold changed the world, and showed the field what was possible with AI for science and medicine, lighting the way for how AI can benefit humanity” [4][6][15].

Jumper was also a key member of Google’s team developing AI coding tools, an area where the company has struggled to sell to businesses, according to former employees [3][4][6]. His move to Anthropic directly strengthens a rival that is already pulling ahead in enterprise AI coding with its Claude Code product. Bloomberg reported that Jumper’s departure “is not helping Google” and that “OpenAI and Anthropic are increasingly the dominant frontier firms in the US and seem to be pulling away from models and coding tools from Google, Meta, and xAI” [8][10].

The Transformer Diaspora: Foundational Researchers Who Left Earlier

The departures of Shazeer and Jumper are the latest in a long line of exits by the very researchers who invented the Transformer architecture. All eight authors of “Attention Is All You Need” have now left Google [2][14]. Key figures include:

• Ilya Sutskever: Co-inventor of the Transformer, left Google Brain in 2015 to co-found OpenAI. He later left OpenAI in 2024 to found Safe Superintelligence Inc. (SSI) [1][2].
• Jakob Uszkoreit: Co-author of the Transformer paper, left Google in 2021 to co-found Inceptive, a biotech AI startup that recently signed a $2 billion collaboration with Alnylam Pharmaceuticals [12][13][16].
• Aidan Gomez: Co-author, left Google Brain in 2019 to found Cohere, an enterprise AI company positioning itself as a sovereign alternative to US tech giants [17].
• Ashish Vaswani and Niki Parmar: Co-authors (Vaswani is first author), left Google in 2022 to co-found Adept AI, then left Adept in 2023 to found Essential AI, focused on enterprise AI [1][2].
• Geoffrey Hinton: Often called a “godfather of deep learning,” left Google in 2023 to speak freely about AI risks, becoming a prominent public voice on AI safety [1][2].

The fact that the creators of the most important AI architecture of the decade have all departed Google is a symbolic and practical loss. It signals that the company has been unable to retain the very people whose work now powers the entire industry.

Other Notable Departures and the Broader Pattern

Beyond the Transformer authors, other departures illustrate a wider retention problem. Daniel De Freitas, co-founder of Character.AI with Shazeer, left Google in 2021 and returned in 2024 as part of the $2.7 billion deal, though his current status is uncertain [2][5][14]. Matt Lowrie, a test engineer with nearly 19 years at Google, quit in November 2025, citing pressure to adopt AI too quickly and a feeling that he had “aged out” of the company [4][5]. Aashna Doshi, a software engineer, left in May 2026 to launch an AI startup called Bounty, stating that “financial security is comfortable, but it can also be a trap” [18].

Analysts have framed these exits as a structural disadvantage for Google. D.A. Davidson analyst Gil Luria noted: “There is so much demand for limited AI research talent that the frontier AI research labs are willing to do whatever it takes to add them. This puts OpenAI and Anthropic at an advantage over large companies like Google because they can promise less bureaucracy and a more focused effort on pursuing Superintelligence” [7][8]. Luria added: “Google is losing the war for talent at the frontier of AI. Google had the state-of-the-art model for a few weeks last year which helped it get credit as an AI winner but has fallen off since, and these departures may mean it is falling behind” [8][9].

Impact on Alphabet’s Key AI Products

Gemini: Leadership Vacuum and Competitive Pressure

Noam Shazeer’s departure directly destabilizes the Gemini product line. As co-lead, he was responsible for the strategic direction and technical roadmap of Google’s flagship AI model. His exit comes at a critical juncture: Google recently unveiled Gemini Spark (an always-on autonomous agent) and Gemini Omni (a video-based “super-agent”) at I/O 2026, alongside Gemini 3.5 Flash and Gemini 3.1 Pro models [27][33]. The company also announced a major overhaul of Search, centering on an AI-first experience powered by Gemini 3.5 models, with AI Mode as the default interface [33]. Losing the co-lead of these initiatives risks slowing execution and decision-making at a time when competitors are iterating rapidly.

Despite these challenges, Gemini has shown impressive user growth, reaching 900 million monthly users, up from about 400 million the previous year [27][33]. Google has also leveraged its infrastructure cost advantages, with CEO Sundar Pichai noting that using a mix of Flash and other frontier models could save over $1 billion a year for top Google Cloud customers [29]. However, a partial outage on June 10, 2026, affecting the Gemini app across web, Android, iOS, and Workspace, highlighted ongoing reliability concerns [14].

Google Cloud AI and Vertex AI: Coding Tools Struggle and Security Vulnerabilities

John Jumper’s departure is a blow to Google’s AI coding tool efforts, an area where the company has already struggled to gain traction against GitHub Copilot (Microsoft) and Claude Code (Anthropic) [3][4][6]. Google Cloud’s Vertex AI platform also suffered a critical security vulnerability in early 2026. Dubbed “Pickle in the Middle” by Palo Alto Networks Unit 42, the flaw allowed attackers to hijack machine learning model uploads and execute arbitrary code inside Google’s serving infrastructure [14]. The vulnerability stemmed from predictable Cloud Storage bucket names in the Vertex AI SDK for Python. Google patched it in March and April 2026, but it was the second such predictable-bucket-name flaw in Vertex AI that year, raising concerns about the platform’s security posture [14].

Google Cloud revenue added $17.6 billion in the most recent quarter (up 12.3% year-over-year), with an annual run rate exceeding $70 billion, driven by AI demand [4]. Yet the cloud market remains dominated by Amazon (28% share) and Microsoft (21%), with Google at 14% [1]. The talent drain in AI coding and infrastructure could further widen this gap.

Google Search AI Features: AI Overviews and AI Mode Under Regulatory Scrutiny

Google’s AI Overviews now reach over 2.5 billion monthly active users, and AI Mode has over 1 billion monthly users [30][31]. The company has given website owners an opt-out toggle for AI Overviews and AI Mode via Search Console, though Google stated that using this control will not be used as a ranking signal [30]. However, UK regulators (the Competition and Markets Authority) have ordered Google to provide clearer attributions and links to publishers’ content in AI-generated search features and to give publishers an effective opt-out mechanism, ruling that Google cannot penalize opted-out publishers by downranking them in general search results [32]. Google has nine months to comply.

An analysis by Lily Ray found that Google’s AI Overviews increasingly decouple which pages they cite from which brands they recommend, with roughly 69% of cases showing a site’s self-promotional listicle cited but omitted from the recommendation [19]. This suggests that citation counts may be a misleading proxy for AI-driven visibility. Meanwhile, AI Overviews are now triggered on 75% of pharma searches, up from 50% the previous year, significantly impacting organic traffic patterns [31].

The talent departures do not directly cause these regulatory and design challenges, but they weaken the internal expertise needed to navigate them while simultaneously innovating in search AI.

Alphabet’s Talent Retention Efforts: What Went Wrong?

Compensation and Equity Structures

Alphabet has not been stingy with compensation. The company’s CFO Anat Ashkenazi received $31.3 million in 2025, the second-highest among S&P 500 CFOs [34]. Google has also used massive acqui-hire deals, such as the $2.7 billion Character.AI transaction, to bring back key talent like Shazeer [2][5][14]. However, the equity upside available at pre-IPO AI startups dwarfs what a mature public company like Alphabet can offer. Financial planners report that one Anthropic client with just three years of tenure already has $40 million in vested equity and $30 million more vesting [35]. OpenAI employees hold Profit Participation Units (PPUs) that will convert to regular shares in the upcoming IPO, with some 2022 and 2023 equity grants now valued at over $50 million [35]. Nvidia, another magnet for AI talent, pays base salaries of up to $471,500 for distinguished AI algorithm engineers, with stock awards adding substantially more [36].

Alphabet’s stock-based compensation, while generous by conventional standards, cannot match the life-changing wealth creation potential of joining a high-growth AI lab on the cusp of an IPO. This structural disadvantage is compounded by the fact that Alphabet’s share price has been under pressure from AI spending concerns, making equity grants less attractive relative to rivals whose valuations are soaring.

Research Culture, Bureaucracy, and the DeepMind-Brain Merger

The 2023 merger of Google Brain and DeepMind into Google DeepMind was intended to consolidate AI research efforts, but it appears to have introduced or preserved bureaucratic friction. Analyst Gil Luria explicitly cited “less bureaucracy” as a key advantage that OpenAI and Anthropic hold over Google [7][8]. Former employees have described a culture where internal politics and slow decision-making hinder the kind of rapid iteration that characterizes frontier AI labs.

Google DeepMind has attempted to maintain a strong research culture, publishing a 35-page technical road map in June 2026 outlining a security framework to protect against rogue AI agents [21]. The framework borrows from cybersecurity insider-threat prevention and includes dynamic access control, continuous monitoring of agent behavior, and a threat taxonomy called TRAIT&R [21]. While this demonstrates sophisticated thinking, it also reflects the complexity of operating within a large, risk-averse organization.

Compute Resources and Infrastructure Spending

Alphabet has committed staggering sums to AI infrastructure. The company has raised $141 billion in debt and equity since October 2025, with projected AI capital expenditure of approximately $190 billion for 2026 [8][10]. In June 2026, Google signed a $30 billion deal with SpaceX to rent AI compute, paying $920 million per month for access to approximately 110,000 NVIDIA GPUs [38]. A Google Cloud spokesperson said the deal provides “bridge capacity to meet surging customer demand for our agent platform, Gemini Enterprise, which has been even higher than we expected” [38].

Google also enjoys a significant internal cost advantage, paying around 50% less (and possibly as much as 75% less) for its internal AI compute than rivals because it uses its own TPU chips and sources components directly from manufacturers [29]. This full-stack ownership—chips, data centers, cloud, models, and applications—should be a powerful retention tool for researchers who need massive compute resources. Yet the departures suggest that compute abundance alone is not enough to retain top talent if the research environment and equity incentives are perceived as inferior.

Organizational Restructuring and Layoffs

In June 2026, Google conducted a fresh round of layoffs within its Cloud business, affecting teams including its Threat Intelligence Group and Mandiant, the cybersecurity firm acquired for $5.4 billion in 2022 [13]. While Google framed the cuts as routine organizational restructuring, the inclusion of strategically important security teams signals that few parts of the company are immune from resource reallocation toward AI. Such layoffs can damage morale and create uncertainty, making it harder to retain AI talent who may already be weighing offers from more stable or mission-focused competitors.

Google’s approved H-1B visa hires also fell from 5,100 to 2,200 year-over-year, a 64% decline, while Anthropic, OpenAI, and Nvidia increased their H-1B certifications [19]. This suggests that Google is not only losing existing talent but also becoming less aggressive in recruiting foreign AI researchers, even as rivals pay the new $100,000 visa fee without hesitation [19].

How OpenAI and Microsoft Are Winning the Talent War

OpenAI: Equity Upside, Mission Focus, and Aggressive Hiring

OpenAI’s talent strategy rests on three pillars: enormous equity upside, a focused mission to achieve artificial general intelligence (AGI), and aggressive recruitment of star researchers. The company’s confidential IPO filing in June 2026, targeting a valuation of up to $1 trillion, has made its PPUs extraordinarily valuable [13][20]. Employees who joined in 2022 or 2023 are sitting on equity worth tens of millions of dollars, and the impending public listing creates a tangible wealth event that Google cannot replicate [35].

OpenAI’s mission-driven culture is a powerful draw. CEO Sam Altman has set goals of running an automated AI research intern on hundreds of thousands of chips by September 2026 and a true automated AI researcher by March 2028 [4][5]. The company is willing to pay top dollar for safety researchers, offering up to $445,000 in base salary for a role on its Preparedness team focused on recursive self-improvement risks [4][5][6]. OpenAI has also hired Dean Ball, a former White House advisor, to lead a new “Strategic Futures” team on catastrophic risk management and government relations [14][22].

The recruitment of Noam Shazeer is the most visible coup. By bringing in the co-creator of the Transformer and former Gemini co-lead, OpenAI not only gains deep architectural expertise but also deals a psychological blow to Google [2][14][20]. OpenAI’s H-1B visa filings rose from 20 to 63 in Q2 2026, signaling an intensified global talent search [19].

Microsoft: Building a Fourth Frontier Lab with Mustafa Suleyman

Microsoft has transformed its AI strategy under Mustafa Suleyman, the former DeepMind co-founder and Inflection AI CEO who now leads Microsoft AI. At Microsoft Build 2026, Suleyman declared his ambition to make Microsoft one of the top four AI labs alongside Google DeepMind, OpenAI, and Anthropic [9]. He revealed that a contractual change with OpenAI in late 2025 formally granted Microsoft the authority to pursue “superintelligence” using its own researchers, data, and custom silicon [11]. Suleyman stated: “We were only sort of set free from our contract with OpenAI about six months ago to formally pursue superintelligence” [11].

The launch of seven in-house MAI models, including the 35-billion-parameter reasoning model MAI-Thinking-1, is the most tangible evidence of this new independence [11][12]. MAI-Thinking-1 scored 97% on the AIME benchmark and 53% on SWE Bench Pro, competitive with Anthropic’s Claude Opus 4.6 and OpenAI’s GPT-5.4 [12]. Suleyman emphasized that the models were trained from scratch on clean, commercially licensed data, addressing enterprise copyright concerns [12]. Microsoft is also building custom Maia 200 silicon, which Suleyman claims is 30% more cost-efficient than Nvidia’s GB200 [11].

Microsoft’s talent strategy leverages its vast financial resources, enterprise relationships, and the promise of building something new. Suleyman himself was hired from Inflection AI, and Microsoft has been actively recruiting from DeepMind and Google [25]. The company’s partnership with Mayo Clinic to build a frontier healthcare AI model demonstrates its ability to offer researchers meaningful, high-impact projects with real-world data [8]. Microsoft’s AI business crossed a $37 billion annual run rate (up 123% year-over-year), providing the financial momentum to support ambitious talent acquisition [3].

What They Offer That Alphabet Doesn’t

The comparative advantages of OpenAI and Microsoft over Alphabet in the talent market can be summarized as follows:

• Equity wealth creation: Pre-IPO equity at OpenAI and Anthropic has generated paper fortunes for early employees, with some holding vested equity worth $40 million or more [35]. Alphabet’s mature stock cannot offer comparable upside.
• Less bureaucracy: Frontier labs promise a focused mission on superintelligence without the corporate layers of a trillion-dollar conglomerate [7][8].
• Faster iteration: Anthropic shipped Claude Opus 4.8 just 41 days after Opus 4.7 [29]. Microsoft built and launched seven new models in a single event [11]. Google’s product cycles, while accelerating, still reflect the complexity of integrating AI across Search, Cloud, Android, and Workspace.
• Mission clarity: OpenAI’s AGI mission and Microsoft’s “Humanist Superintelligence” framing provide a compelling narrative that contrasts with Google’s more diffuse “organize the world’s information” mandate [11].
• Compute independence: Microsoft is building its own silicon and has the largest GPU purchases, while OpenAI has a planned $600 billion compute infrastructure spend by 2030 [11][13]. Google’s compute advantage from TPUs is real, but it is tied to Google’s internal ecosystem, which may feel restrictive to researchers who want platform flexibility.

The Broader Competitive Landscape

Market Share Shifts: Cloud, Consumer AI, and Enterprise

The talent asymmetry is beginning to manifest in market share data. In cloud infrastructure, Google holds 14% market share, behind Amazon (28%) and Microsoft (21%) [1]. While Google Cloud’s AI-driven revenue is growing, the gap with Microsoft Azure—which benefits from OpenAI’s models and Microsoft’s own MAI family—may widen as Microsoft leverages its enterprise relationships and new AI capabilities.

In consumer AI, ChatGPT’s “True Audience Share” (excluding overlap between apps and web) fell below 50% for the first time in March 2026, reaching 46.4% by May, while Gemini grew to 27.7% and Claude to 10.3% [28]. This suggests that Google is gaining ground in consumer adoption, but the trend also reflects the fragmentation of the market rather than a decisive Google victory. ChatGPT still has over 900 million weekly active users, and its paid subscriber base of about 50 million generates significant revenue [1][2].

In enterprise AI, Anthropic has seized a 40% market share lead versus 27% for OpenAI, according to PitchBook [26]. Anthropic’s annualized revenue skyrocketed from $9 billion at the end of 2025 to over $47 billion in May 2026, driven by its agentic coding assistant Claude Code [37]. The company closed a $65 billion Series H round at a $965 billion valuation, overtaking OpenAI as the most valuable private AI startup [3][5]. Google’s enterprise AI offerings, including Vertex AI and Gemini Enterprise, face an uphill battle against this momentum.

Research Output and Model Release Cadence

The pace of model releases has accelerated dramatically across the industry. Anthropic’s Claude Opus 4.8, released on May 28, 2026, introduced Dynamic Workflows allowing swarms of up to 1,000 parallel subagents, a 4x improvement in honesty, and a 3x price cut on Fast Mode [29][34]. It topped GPT-5.5 on the Artificial Analysis Intelligence Index and led on several key benchmarks [29]. Microsoft’s MAI-Thinking-1 demonstrated competitive reasoning performance at lower cost [12]. Google’s Gemini 3.5 Flash and 3.1 Pro models have been well-received, but the company has not consistently held the state-of-the-art crown, and the loss of Shazeer and Jumper may further slow its research cadence.

A Nature Medicine study published in 2026 found that frontier general-purpose LLMs (GPT-5.2, Gemini 3.1 Pro, Claude Opus 4.6) outperformed specialized clinical AI tools on medical benchmarks, with Gemini achieving the highest accuracy on MedQA (97.4%) [30]. This demonstrates that Google’s models remain highly capable, but the competitive gap is narrow and volatile.

The AI Cost Spiral and Infrastructure Arms Race

The entire industry is grappling with an AI cost crisis. Despite falling token prices, usage volume has surged so dramatically that companies are exhausting annual AI budgets within months. Uber depleted its entire AI coding budget for 2026 by April, and Microsoft revoked Claude Code licenses from its developers after just a few months [32]. One company reportedly paid $500 million for Claude because it forgot to set employee limits [2][32]. This cost spiral is forcing enterprises to prioritize efficiency and token control, an area where Google’s full-stack cost advantage could become a decisive competitive weapon [29].

However, the infrastructure arms race is also creating a financial overhang. Combined AI capex by Google, Microsoft, Amazon, and Meta is projected to reach nearly $600 billion in 2026, yet Sequoia’s David Cahn estimates a ~$600 billion annual revenue gap between hyperscaler AI spending and actual AI ecosystem revenue [3][35]. If models become more interchangeable, as Microsoft CEO Satya Nadella suggested in a recent interview, investors may question whether Google’s $190 billion AI capex is building a durable advantage or simply adding pressure to margins [8][10].

Counter-Examples: Signs of Resilience at Alphabet

While the talent exodus is alarming, Alphabet retains significant strengths that could mitigate the damage. Google DeepMind CEO Demis Hassabis remains at the helm and has publicly predicted that AGI could be realized by 2030, signaling continued ambition [23][24]. At the G7 summit in June 2026, Hassabis joined Anthropic CEO Dario Amodei in calling for a U.S.-led international AI coalition, demonstrating that Google DeepMind still commands geopolitical influence [25].

Google’s product momentum is tangible. Gemini has grown to 900 million monthly users, and AI Overviews reach over 2.5 billion users [27][30][33]. The company’s $75 million investment in indie film studio A24 to develop AI filmmaking tools shows a willingness to explore creative applications that could attract talent interested in arts and media [22]. The $30 billion SpaceX compute deal, while ironic given that SpaceX’s xAI competes with Google, ensures that Google has access to the GPU capacity needed to serve surging Gemini Enterprise demand [38].

Alphabet has also continued to invest in Anthropic, committing up to $10 billion more at a $350 billion valuation, which gives it financial exposure to a leading frontier lab even as it loses talent to that same lab [7]. Google’s internal compute cost advantage—paying 50-75% less than rivals for AI compute—is a structural moat that could become increasingly important as the AI cost spiral forces enterprises to seek cheaper inference [29].

Moreover, not all AI talent is leaving. Google’s AI workforce remains one of the largest in the world, and the company continues to publish significant research. The DeepMind team still includes many world-class scientists, and the broader Google AI organization has deep benches in areas like reinforcement learning, computer vision, and robotics. The high-profile departures are disproportionately from the Transformer-era cohort; a new generation of researchers may rise within Google, though the company must first stabilize its leadership and culture.

Conclusion

The departures of Noam Shazeer and John Jumper in June 2026 are the most visible symptoms of a deeper talent retention crisis at Alphabet. The company has lost not only the co-lead of its flagship Gemini models and a Nobel Prize-winning scientist, but also the entire cohort of researchers who invented the Transformer architecture. These exits directly threaten Google’s ability to compete at the frontier of AI development, particularly in coding tools, enterprise AI, and the next generation of autonomous agents.

Alphabet’s retention strategies—massive acqui-hire deals, enormous compute investments, and full-stack infrastructure—have proven insufficient against the lure of pre-IPO equity, mission focus, and bureaucratic freedom offered by OpenAI, Anthropic, and Microsoft. The competitive landscape is shifting: Anthropic has overtaken OpenAI in valuation and enterprise market share, Microsoft is building a self-sufficient frontier lab under Mustafa Suleyman, and Google’s cloud market share remains a distant third.

Yet Alphabet is not without recourse. Its compute cost advantage, massive user base, and deep integration across Search, Android, and Workspace provide distribution and efficiency that pure-play AI labs cannot match. If Google can address its cultural and bureaucratic impediments, restructure compensation to offer more entrepreneurial upside, and stabilize the leadership of its AI product teams, it may yet retain and attract the talent needed to remain a top-tier AI power. The window for such a turnaround is narrowing, and the events of June 2026 have made the stakes unmistakably clear.