Chef's editorials

Microsoft’s Startup Reset: Why Satya Nadella Is Betting on a 33-Year-Old to Save Copilot

by
Tom Cironis
June 29, 2026
For years, Microsoft was viewed as one of the biggest winners of the AI revolution. Its multi-billion-dollar investment in OpenAI gave the company an early lead, while Copilot promised to bring generative AI to hundreds of millions of enterprise users.

Fast forward to today, and the picture looks very different.

ChatGPT continues to dominate consumer AI, Anthropic has become a serious enterprise challenger, and Microsoft is now trying to rediscover the speed and execution of a startup. Leading that transformation is Jacob Andreou, a 33-year-old executive who joined Microsoft just a year ago and has already become one of CEO Satya Nadella‘s most important bets.

 

From Snap to Microsoft’s AI turnaround

Before joining the company, he spent eight years at Snap, where he helped scale the social platform from roughly 80 million to more than 360 million daily active users. He later became a venture partner at Greylock, investing in consumer AI startups before Microsoft recruited him in 2025 to strengthen its AI division.

Only twelve months later, Nadella promoted him to Executive Vice President responsible for Copilot, a remarkably fast rise inside a company with more than 220,000 employees. 

"This is one of the most intensely competitive environments tech has seen in the last 20 years. A six-to-twelve-month roadmap doesn't really exist in the way it used to."

Andreou has brought a noticeably different operating style. Instead of long planning cycles, his teams ship quickly, iterate aggressively, and work in compact engineering groups. One early demonstration of this approach was Copilot Tasks; an AI agent capable of autonomously completing multi-step actions, such as ordering food online. The product reportedly went from concept to working prototype in just two months.

Andreou is known for writing code himself, joining engineers during development, and pushing teams toward rapid execution. Employees describe an environment that increasingly resembles a startup rather than a traditional enterprise software company.

Andreou has merged Microsoft’s previously separate consumer and enterprise Copilot organizations, removed overlapping products, simplified internal decision-making, and introduced usage-based pricing alongside traditional enterprise licensing. His team is also developing a new Copilot “super app” that aims to combine chat, coding assistance, autonomous AI agents, and personal and work data into a single interface. The goal is simple: make Copilot a daily AI workspace rather than just another feature inside Microsoft 365.

 

The challenge is bigger than building good AI

Despite Microsoft’s enormous resources, Copilot has struggled to gain the same momentum as ChatGPT. Only around 4.5% of Microsoft’s roughly 450 million Microsoft 365 users currently pay for Copilot features, while OpenAI continues to dominate mindshare in consumer AI. Investors have also questioned Microsoft’s massive AI infrastructure spending and its dependence on OpenAI, leading to increased pressure on management to demonstrate stronger commercial results.

Recent wins, including large-scale deployments across regulated industries such as healthcare, suggest Microsoft still enjoys enormous trust among enterprise customers. 

 

Startup culture inside a trillion-dollar company

Perhaps the most interesting part of Andreou’s story isn’t the technology. It’s the organizational philosophy. Microsoft increasingly appears to believe that competing with AI startups requires behaving more like one. Hackathons have become more frequent. Smaller engineering teams are encouraged to move independently. Product decisions happen faster. Bureaucracy is being reduced.

Andreou has openly championed the idea of “10x developers”; engineers capable of dramatically outperforming their peers through technical excellence and AI-assisted workflows. The approach has produced faster product releases, but it has also drawn criticism. Some current and former employees describe long working hours, burnout, and growing pressure to keep pace with rivals such as OpenAI and Anthropic. It’s the classic startup trade-off: speed versus sustainability.

 

Microsoft’s next chapter

The AI race is nowadays more about building products that become indispensable inside companies while maintaining security, compliance, and trust—areas where Microsoft believes it still holds a competitive advantage. Andreou has outlined three priorities for Copilot: (1) Build the best AI chat experience, (2) Deliver leading AI models without falling behind competitors, and (3) Offer a trusted platform capable of integrating multiple AI models securely.

Whether that strategy succeeds remains uncertain. What is clear is that Microsoft’s AI future increasingly depends not on another billion-dollar investment, but on whether one young product leader can bring startup execution to one of the world’s largest technology companies.

For founders, there’s an interesting lesson here: even trillion-dollar companies eventually discover that bureaucracy is their biggest competitor. Sometimes, the fastest way forward is to start thinking like a startup again.

Subscribe to our newsletter

Join our newsletter to get the best hottest startup tapas.

Popular posts

Ment2Grow Accelerator Launches Its 9th Cohort – Helping 15 Czech B2B Startups Grow Faster

September 25, 2026

Swiss robotics startup Embodied AI Raises Funding to Bring Learning Robots to European Factories, among investors Czech Look AI Ventures

September 25, 2026

Czech Republic Launches First Technology Transfer Fund, Targeting €50 Million for AI and Deep Tech

September 24, 2026

Related posts

The IVC founder talks about getting into the defense game and the future of Europe.
by
Fergus O'Sullivan
As AI moves from screens into the physical world, one problem is becoming increasingly difficult to ignore: machines cannot learn from data they cannot easily collect. Domonkos Haffner, Founding Generative AI Engineer and AI Architect at DiffuseDrive, is working on exactly that challenge. His team uses generative AI to create targeted synthetic data for the rare, dangerous and expensive scenarios that computer-vision systems struggle to capture in the real world. In this conversation, Domonkos explains why synthetic data is not about generating endless images, how to measure whether artificial data actually improves AI, and what it takes to build reliable AI systems in a field where the technology can change every few months. That’s why we sat down with Domonkos Haffner, Founding Generative AI Engineer & AI Architect at DiffuseDrive, and also one of the industry leaders.
by
Jakob Ulrych
Investing into Ukraine is the topic of the day among VCs, but what exactly draws investors to the war-torn country?
by
Fergus O'Sullivan
As AI moves from screens into cars, robots, drones, factories and defence systems, the data problem is changing. The next generation of physical AI may depend less on collecting more data, and more on generating precisely the data that does not yet exist. At DiffuseDrive, Daniel Bin Schmid, Domonkos Haffner and Gabor Vecsei are working on different parts of that problem, from controllable generative models to production-grade, air-gapped AI infrastructure.
by
Jakob Ulrych