Microsoft has lured more than two dozen engineers and researchers away from Google’s DeepMind in a dramatic escalation of the artificial intelligence talent war, according to a new report from The Wall Street Journal. The hiring blitz is being orchestrated by Mustafa Suleyman, the DeepMind co-founder who now heads Microsoft AI, and it signals a fierce new front in the battle to dominate the next wave of AI-powered products.

Suleyman is personally cold-calling and messaging former colleagues, dangling the allure of a startup-like culture, substantial pay hikes, and the chance to build AI systems that directly touch hundreds of millions of Microsoft users through Copilot, Windows, Azure, and beyond. More than two dozen current or recent DeepMind employees have accepted the offer — and at least another dozen are rumored to be contemplating the jump.

The talent drain from DeepMind is not just a story about corporate poaching; it’s a strategic maneuver that could reshape the AI landscape for years. With key architects of Google’s Gemini assistant now ensconced inside Microsoft, the Redmond giant is assembling a dream team that rivals the best research labs in the world.

Inside the Recruitment Blitzkrieg

Suleyman’s approach is methodical and personal. Rather than relying on generic LinkedIn messages or third-party recruiters, he is tapping his own network built during his years at DeepMind. Insiders describe a carefully choreographed courtship that begins with a simple text or a WhatsApp call.

“He tells them Microsoft wants to win, and he needs the best people to do it,” said a source familiar with the outreach. “There’s a sense of urgency and excitement — like in the early days of DeepMind, but backed by Microsoft’s immense resources.”

The pitch, corroborated by several new hires, leans heavily on three pillars:

  • Unprecedented agility: Suleyman has been granted a degree of independence rarely seen inside a tech giant the size of Microsoft. His division operates almost like a startup-within-the-enterprise, with its own culture, rapid decision-making, and minimal red tape. For researchers frustrated by what some describe as Google’s increasingly bureaucratic processes, this is a potent draw.
  • Compensation that stings: While Microsoft is not handing out the nine-figure “superstar” packages that OpenAI or Anthropic have dangled in front of a handful of researchers, it is offering substantially larger salary and equity packages than what DeepMind typically provides. Recruiters familiar with the moves say base salaries for senior roles are jumping 20-40%, and equity grants can swell total compensation into seven figures annually. For many, that’s hard to ignore.
  • Proximity to product: At DeepMind, researchers often work on long-term, blue-sky problems with an uncertain path to commercialization. At Microsoft, Suleyman promises they will see their work land in real products — Copilot in Windows 11, Edge, and Office, as well as Azure AI services — within months, not years.

This combination has proven remarkably effective. In a matter of weeks, Microsoft’s AI division has absorbed a cohort of talent that would normally take years to recruit.

The Prize Recruits

Among those who have already made the switch are some of the most respected names in the AI community. Their arrivals have sent shockwaves through the industry and prompted emergency retention meetings at Google.

  • Amar Subramanya, former vice president of engineering at Google and a key architect of several large-scale AI infrastructure projects, has taken the title of corporate vice president at Microsoft AI. His expertise in scaling massive language models will be critical as Microsoft races to integrate AI across its product line.
  • Adam Sadovsky, a distinguished engineer at DeepMind known for pushing the boundaries of reinforcement learning applied to real-world systems, is now also a corporate vice president at Microsoft AI. Colleagues describe him as a “force multiplier” who can bridge research and production with rare ease.
  • Sonal Gupta, previously an engineering lead at DeepMind where she helped steer the development of next-generation assistants, has joined as a principal member of the technical staff. Her work on conversational AI and multi-modal systems is expected to directly influence Copilot’s evolution.
  • Jonas Rothfuss, a research scientist whose work on few-shot learning and adaptive models has been widely published, moved into a technical staff role. His hiring signals Microsoft’s intent to stay at the cutting edge of foundational research, not just application.

These four are just the publicly confirmed tip of the iceberg. Multiple current and former DeepMind employees told The Wall Street Journal that at least two dozen people have already moved, with many more in talks. The exits span senior researchers, engineers, product managers, and even operations staff — a full-fledged brain drain that Google is scrambling to stem.

The Suleyman Factor

Mustafa Suleyman’s own journey is central to this story. After co-founding DeepMind in 2010 with Demis Hassabis and Shane Legg, he led the company’s applied AI division until 2019, when he moved to Google as vice president of AI policy. He left Google in 2022 to start Inflection AI, a startup that built a personal AI called Pi. But in March 2024, Microsoft hired him and much of Inflection’s team in a $650 million licensing deal, effectively bringing the startup in-house while avoiding the regulatory scrutiny of a full acquisition.

That move gave Suleyman a mandate to build Microsoft’s AI future. Now, he’s using his DeepMind pedigree and personal charisma to raid his former stomping grounds. “It’s an open secret inside DeepMind that if you want to work on cutting-edge AI that ships fast, Mustafa’s new team is the place to be,” said a current DeepMind researcher who requested anonymity because they were not authorized to speak to the press.

The exodus has struck a nerve at Google, where retention has become a top priority. In recent months, the company has loosened some internal policies, accelerated promotion cycles, and even handed out special stock grants to keep top researchers from leaving. But for those who have already departed, the pull of Suleyman’s vision and Microsoft’s scale proved too strong.

Microsoft’s Broader AI Ambitions

This hiring spree does not exist in a vacuum. Microsoft has been on an AI talent-buying binge for over a year. After its multi-billion-dollar investment in OpenAI, the company integrated ChatGPT into Bing, Windows, and Office, rebranding everything under the Copilot umbrella. But Copilot’s early versions, while promising, have faced criticism for occasional errors, latency, and limited multimodal capabilities.

By importing DeepMind’s finest, Microsoft aims to close the gap with — and perhaps surpass — Google’s Gemini and OpenAI’s GPT models. The new hires are expected to accelerate development in several key areas:

  • Copilot’s conversational abilities: Gupta’s expertise in multi-turn dialogue systems could make Copilot more natural and context-aware.
  • Infrastructure and scale: Subramanya’s background in AI infrastructure will help Microsoft optimize Azure’s AI services for lower latency and higher throughput at lower cost.
  • Research-to-product pipeline: Sadovsky’s talent for productizing research could shorten the time from paper to practical feature, giving Microsoft a speed advantage.
  • Adaptive learning: Rothfuss’s work on few-shot learning may enable Copilot to personalize itself to individual users with minimal data, a key differentiator in enterprise settings.

Suleyman has been cagey about exact product timelines, but internal sources indicate that a major Copilot update incorporating some of this new talent’s work could arrive as early as the second half of this year. That update is expected to bring tighter Windows integration, the ability to assist more deeply with creative and coding tasks, and a persistent memory that learns from user habits over time.

The AI Talent War Goes Nuclear

The recruitment battle between Microsoft and Google is emblematic of a broader talent crunch in AI. Global demand for machine learning engineers and researchers far outstrips supply, and the stakes have never been higher. A single star hire can shift the competitive balance — and a mass defection can leave a company scrambling.

Compensation has spiraled into absurd territory for the absolute top tier. While the nine-figure offers remain rare, millions of dollars in total compensation for senior researchers are now common across the giants. Equity packages often vest over several years and are tied to aggressive performance milestones, essentially turning hires into founders within a corporate shell.

But money alone isn’t enough. Culture and autonomy have become decisive factors. Many top AI researchers disdain the bureaucracy of large tech companies and prefer the fast-paced, mission-driven feel of a startup. Microsoft’s play is to offer the best of both worlds: the resources and stability of a $3 trillion company with the culture of a nimble startup. Suleyman’s division is reportedly free from the traditional Microsoft performance review cycle and can set its own engineering practices.

Google, for its part, is trying to fight back. The company has been merging its research arms (Google Brain and DeepMind) into a single unit to reduce internal competition and overhead, and has given DeepMind’s leadership more direct product oversight. But this very reorganization may have contributed to the exodus: some researchers feel the new structure is less focused on fundamental research and more on immediate product needs, eroding the lab’s original ethos.

What This Means for Windows Users

For the millions of people who use Windows, Edge, and Microsoft 365 every day, this talent infusion could translate into tangible improvements soon. Copilot is already being positioned as the central AI assistant across Microsoft’s ecosystem, but it still feels like a work in progress. With DeepMind veterans now steering engineering, expect Copilot to become more capable, more reliable, and more seamlessly embedded into the operating system.

Imagine a Copilot that not only answers questions but proactively organizes your files, drafts entire project proposals based on a few bullet points, and manages your calendar with near-human intuition. Or one that can look at a screenshot of an error message and walk you through the fix step by step. These are the kinds of experiences that the new hires are uniquely qualified to build.

Moreover, Windows itself stands to benefit. Microsoft has been adding AI features to Windows 11 at a steady clip: AI-powered Studio Effects for video calls, voice clarity enhancements, and even a “Recall” feature that uses AI to create a searchable timeline of everything you do on your PC. The DeepMind team brings deep expertise in systems that learn from visual and linguistic cues, which could supercharge these local AI features and make Windows the most intelligent OS on the market.

Ethical and Competitive Considerations

Such aggressive poaching inevitably raises ethical questions. Is Microsoft simply buying its way to dominance by stripping a competitor of its most valuable asset? And what does this concentration of power mean for the broader AI ecosystem?

Regulators have been paying close attention. Microsoft’s deal with Inflection AI already drew scrutiny from the Federal Trade Commission over whether it circumvented merger review rules. The DeepMind hiring spree is likely to fuel further debate about how tech giants compete for talent and whether antitrust laws need to be updated for the AI age.

There is also a risk of brain drain leading to a de facto duopoly. If only Microsoft and a few others can afford to retain top AI researchers, innovation could slow as talent becomes locked into a handful of giant companies. However, others argue that competition for talent spurs investment in education and training, ultimately growing the overall talent pool.

For now, Suleyman and Microsoft show no signs of slowing down. The company is actively advertising dozens of new AI roles on its careers page, many based in London — DeepMind’s backyard — suggesting that this is just the opening salvo in a long war for AI supremacy.

Forward Look

Microsoft’s DeepMind poaching campaign is a defining moment in the AI talent wars. It illustrates that even the most storied research institutions are vulnerable when a charismatic leader offers a compelling vision, generous pay, and the chance to see their work touch billions of users.

As Suleyman himself said in a recent internal memo: “We are building something genuinely new — a team that can move at the speed of a startup but with the reach of the world’s most ubiquitous software platform. The people joining us aren’t just changing jobs; they’re changing the trajectory of AI.” Whether that trajectory leads to a Microsoft-dominated future or merely provokes a furious comeback from Google remains to be seen. But one thing is clear: the battle for AI’s brightest minds is only heating up.

The next time you ask Copilot to summarize a document or generate an image, you might just be interacting with technology built by the very people who once made Google’s AI sing. And for Windows enthusiasts, that’s a development worth watching very closely.