Microsoft AI CEO says there will be a big price to pay to stay in the AI game.
Mustafa Suleiman said on Wednesday's episode of the Moonshot with Peter Diamandis podcast that competing on the AI front over the next five to 10 years will cost “hundreds of billions of dollars.”
“Not to mention the price we are paying for individual researchers and technical staff,” he added.
Suleiman likened Microsoft to a “modern construction company” with hundreds of thousands of employees building gigawatts of CPUs and AI accelerators.
The scale of investment required is huge and “there are clear structural advantages to being within a large company,” he said.
Microsoft, which has a market capitalization of $3.54 trillion, generated revenue of $77.7 billion in the quarter that ended in September, beating analyst expectations.
Suleiman said his mission is to make Microsoft “self-sufficient” in developing frontier models and to build “absolutely world-class superintelligence teams.”
“We are absolutely moving towards the frontier,” Suleiman said. “We want to build the best superintelligent model and the safest superintelligent model in the world.”
Suleiman said last month that his team is trying to build a “humanistic superintelligence”, one that is aligned with humanity's interests.
It's “hard to say” whether startups will be able to compete with Big Tech because of the high costs of catching up with AI, Suleiman said on the podcast.
“That ambiguity is what causes the valuations to bubble up,” he said. “If there's a sudden explosion of intelligence, a lot of people can get there at the same time.”
High cost of AI
Microsoft is among the tech giants pursuing artificial general intelligence, or superintelligence, and executives across the industry are outspoken about how much that pursuit will cost.
Artificial general intelligence (AGI) refers to AI systems that rival human intelligence in most tasks. Superintelligence goes one step further and is a system that exceeds human capabilities.
Meta CEO Mark Zuckerberg said in September that he would rather risk “misusing hundreds of billions” than fall behind in superintelligence.
If superintelligence arrives sooner than expected and companies move too slowly, it will “move us out of what I think will be the most important technology that enables the most novel products and innovations and value creation and history,” Zuckerberg said.
Billions of dollars are also being poured into AI data centers. In recent months, Big Tech companies such as Microsoft, Meta, Google, and Amazon have increased spending on cloud and computing infrastructure to train and run frontier models.
