Sundar Pichai, Google CEO: "The Risk of Under-Investment in AI Is High"
Google has reportedly set a goal to double the capacity of its related services every six months to meet the growing demand for artificial intelligence (AI).
According to a report by CNBC, a U.S. business media outlet, on November 21 (local time), Google Cloud Vice President Ghaamin Bahdad stated in an 'AI Infrastructure' report presented at a company-wide meeting on November 6, "We now need to double our (computing capacity) every six months," adding, "In the next four to five years, it needs to be 1,000 times larger."
In the report obtained by CNBC, Vice President Bahdad said, "Essentially, we need to provide 1,000 times greater capacity, computing, and networking with the same cost and the same power and energy consumption," and added, "It won't be easy, but we will achieve this goal through collaboration and joint design."
He explained that, in addition to expanding physical infrastructure such as data centers, Google can also enhance its AI processing capabilities by upgrading its proprietary AI chips. He further emphasized that increased spending on infrastructure development is aimed at providing customers with more stable, superior, and scalable infrastructure.
At the same meeting, Google CEO Sundar Pichai addressed the so-called 'AI bubble' controversy and, in response to an employee's concern about overinvestment, said, "During times like these, the risk of underinvestment is very high."
Pichai pointed out, "Cloud performance has been remarkably strong, but if we had more computing resources, the numbers could have been even higher." He went on to cite the company's financial soundness, emphasizing, "We are in a position to withstand mistakes better than other companies."
However, he commented on next year's AI market, saying, "(Competition) will be intense," and added, "There will definitely be ups and downs."
Recently, Google unveiled its new AI model 'Gemini 3,' which achieved top scores in various benchmarks. It also introduced 'Nano Banana Pro,' an image generation and editing tool based on this model.
Google is actively investing in data center expansion and has also launched its proprietary AI chip called 'Ironwood,' known as a Tensor Processing Unit (TPU).
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
