The Wall Street Journal previously reported that the real (inflation-adjusted) amount of money pledged by top tech companies over the past three years to invest in artificial intelligence (AI) data centers, chips, and energy has exceeded the four-decade cost of building the U.S. interstate highway system. Analysts estimate that the useful life of most AI processors is about three to five years.
Investment costs
Computing power development costs
Sequoia Capital partner David Cahn noted that, based on estimated AI infrastructure investment in 2023 and 2024 alone, approximately $800 billion in AI product revenue would be needed to generate a reasonable return on investment.
OpenAI CEO Altman previously stated that OpenAI's vision is to add 1 gigawatt of computing infrastructure per week using standardized processes.
The Financial Times reported on Tuesday that OpenAI executives estimate that the cost of developing 1 gigawatt of computing power is approximately $50 billion, with approximately two-thirds going to purchasing chips and network equipment, and the remainder to cover land acquisition, data center development, and other expenses.
Capital expenditures
David Crawford, chairman of Bain's global technology practice, said on Tuesday that AI will increasingly strain global supply chains, estimating that by 2030, tech executives will need to deploy approximately $500 billion in capital expenditures and generate approximately $2 trillion in new revenue to meet demand while achieving profitability.
Meta Platforms Inc. CFO Susan Li stated at a Goldman Sachs technology conference that the $600 billion mentioned by Meta CEO Mark Zuckerberg at the White House on September 4th refers to Meta's total spending in the United States from this year to 2028, encompassing all of Meta's U.S. data center infrastructure projects and all investments supporting its U.S. operations.
In his annual letter to shareholders in March, BlackRock CEO Mark Fink stated that global new infrastructure investment demand is estimated to reach $68 trillion between 2024 and 2040, and that the construction cost of an AI data center could range from $40 billion to $50 billion.
Revenue Return
Revenue needs
Sequoia Capital partner David Cahn noted that, based on estimated AI infrastructure investments in 2023 and 2024 alone, approximately $800 billion in AI product revenue would be needed to generate a healthy return on investment.
David Crawford, chairman of Bain's global technology practice, estimated on Tuesday that tech executives will need to find approximately $2 trillion in new revenue by 2030 to meet demand and achieve profitability.
Paid Monetization
Business Insider reported that Goldman Sachs internet industry analyst Eric Sheridan recently stated in an interview that nearly every executive he spoke with at a Goldman Sachs technology conference said that demand for AI is outstripping their ability to provide intelligent computing power. Sheridan emphasized that unlike during the dot-com bubble of the 1990s, consumers and businesses are now paying for AI services.
CNBC reported that Google Cloud CEO Thomas Kurian recently stated that Google Cloud has already generated billions of dollars from AI.
MarketWatch reported that MoffettNathanson analyst Michael Nathanson stated on Thursday that Alphabet is becoming the company most capable of monetizing and scaling the generative AI opportunity. Currently the world's fourth-largest company by market capitalization, Alphabet is poised to surpass Microsoft, Apple, and Nvidia to take the top spot.
Source: Content from moneyDJ
