London: Google DeepMind chief executive Demis Hassabis has revealed that a simple but critical challenge drove the decision to sell DeepMind to Google in 2014. The company had ambitious plans to build artificial general intelligence, but it lacked the financial resources needed to turn that vision into reality.
Speaking in a recent interview, Hassabis said DeepMind had estimated that developing artificial general intelligence, often called AGI, could require around $7 billion in investment. At that time, however, the company struggled to raise even $10 million in funding. The gap between its ambitions and its financial resources convinced the founders that they needed a partner with deep pockets and a long term commitment to research.
Hassabis explained that DeepMind was still a young research company with no commercial products and very little revenue. Even though it had attracted some investors, the scale of computing power and engineering talent needed to continue its work was far beyond what a small independent startup could afford.
The turning point came after DeepMind published groundbreaking research showing that its artificial intelligence system could learn to play Atari video games using only the images on the screen. Unlike earlier computer programs that relied on human written rules, the AI learned through trial and error, demonstrating that a single system could master many different games using the same learning method.
The research attracted immediate attention across Silicon Valley. According to Hassabis, major technology companies quickly recognized the importance of the breakthrough and began trying to recruit DeepMind's researchers. Companies including Google, Facebook and entrepreneur Elon Musk all showed interest in the company and its team.
Google eventually acquired DeepMind in 2014 in a deal reported to be worth around $500 million. Hassabis said Google stood out because it was willing to support DeepMind's long term scientific goals instead of demanding quick commercial returns. The company also offered access to powerful computing infrastructure, world class engineering talent and the financial backing needed to continue pursuing advanced AI research.
More than a decade later, that decision has helped shape Google's position in the global artificial intelligence race. DeepMind has gone on to produce several major scientific and technological breakthroughs. Its AlphaGo program defeated world champion Lee Sedol in the ancient board game Go, while AlphaFold transformed biological research by predicting the three dimensional structures of proteins with remarkable accuracy.
In recent years, DeepMind has also become central to Google's efforts to compete with OpenAI and other leading AI companies. In 2023, Google merged its DeepMind and Google Brain research teams into a single organization called Google DeepMind, placing Hassabis in charge of the company's overall AI research strategy.
The combined team has developed several of Google's flagship AI products, including the Gemini family of AI models, Veo for AI generated videos, Imagen for image creation and Lyria for music generation. These technologies are now being integrated into Google's search engine, cloud services and consumer products.
Hassabis has continued to argue that artificial general intelligence has the potential to become one of the most important scientific achievements in human history. He believes advanced AI could accelerate discoveries in medicine, climate science, energy and many other fields, but he has also stressed that governments and technology companies must work together to ensure the technology is developed safely and responsibly.
In recent interviews, he has called for stronger international cooperation on AI governance and independent testing of the world's most advanced AI systems before they are widely deployed. He believes there is still an opportunity to establish global standards that encourage innovation while reducing potential risks.
The interview also highlights how dramatically the economics of artificial intelligence have changed. What once appeared to be an enormous funding requirement has become relatively modest as leading technology companies now invest tens of billions of dollars each year in AI chips, data centers and research. The intense competition among Google, OpenAI, Meta, Anthropic and other companies has made access to computing power and financial resources more important than ever.
Looking back, Hassabis believes DeepMind's decision to join Google was essential for achieving its long term mission. While the company gave up its independence, it gained the resources needed to pursue some of the most ambitious goals in artificial intelligence, helping position Google at the forefront of one of the fastest moving technological revolutions in the world.