But the initial capital cost of open AI is huge.
The explosion of generative AI has taken the world by storm, but one question rarely arises. The question is, who can afford it?
According to industry media The Information, OpenAI squandered about $540 million (roughly R10.5 billion) in developing ChatGPT last year and said it needed $100 billion to achieve its ambitions.
“We will be the most capital-intensive startup in the history of Silicon Valley,” OpenAI founder Sam Altman recently said at a panel discussion.
And when asked how much their AI adventure will cost, Microsoft, which has poured billions of dollars into OpenAI, vows to keep an eye on revenue.
Building something on par with what OpenAI, Microsoft, and Google have to offer will require spectacular investments in cutting-edge chips and the hiring of award-winning researchers.
Independent analyst Jack Gold said, “People don’t realize that doing a fair amount of AI processing like ChatGPT requires an enormous amount of processing power. And these models training can cost tens of millions of dollars.”
“How many companies can really afford to buy 10,000 Nvidia H100 systems at tens of thousands of dollars each?” asked Gold.
The answer is almost no one. In the tech industry, if you can’t build infrastructure, you rent it. This is something companies are already doing at scale by outsourcing their computing needs to his AWS from Microsoft, Google and Amazon.
And with the advent of generative AI, experts warn that the same players will be in the driver’s seat with an even deeper reliance on cloud computing and tech giants.
“Very underrated”
The unpredictable costs of cloud computing are “a hugely underestimated problem for many companies,” said Stefan Sigg, chief product officer at Software AG, which makes software for the enterprise.
Sig likened the cost of the cloud to the cost of electricity, and said it would be a “huge surprise” for obscure companies to rush to develop technologies, including AI, and force engineers to inflate their bills.
Microsoft’s flagship cloud service is Azure, and some observers believe that the giant’s full bet on AI is actually to protect Azure’s success and future-proof the cash cow. I believe there is.
Azure has been an unattractive mainstay for the giant for years, generating huge profits but never making headlines like the consumer-facing iPhone and social media.
For Microsoft, “the golden goose is monetizing the cloud with Azure. Successful bets on AI could create a $20B, $30B, $40B annual opportunity down the road.” because we’re talking about it,” said Dan Ives of Wedbush Securities.
Microsoft CEO Satya Nadella claims generative AI is “quickly moving in the right direction.”
Mr. Nadella, who is deeply respected on Wall Street, will have six to nine months to prove his bet is a winner, Mr. Ives predicted.
Microsoft acknowledges the risks, but it needs to “lead the wave” when it comes to AI, CFO Amy Hood told analysts earlier this month.
“We charge for those AI capabilities and ultimately we get an operating profit,” she said.
collapsed
Even if the company Bill Gates founded is profitable, it’s just passing the cost of AI on to its customers.
From Main Street to the Fortune 500, relying on AI-powered systems is costly, and businesses and investors are looking at alternatives to at least cut their bills.
Spectro Cloud CEO Tenry Fu said, “AI training, GPT training will become a very important cloud service in the future.”
His company, like many others in this space, helps companies optimize cloud technology to save money.
“But after training, companies will be able to bring the models back for real AI applications,” he added, hoping to reduce their reliance on cloud giants.
Regulators hope that they will continue to be able to respond rather than leave the responsibility to the big companies and impose conditions on the smaller ones.
FTC Chairman Rina Khan told CNBC, “Law enforcement officers need to ensure that opportunities and opportunities for competition … are not squashed by incumbents.”
But it may be too late, at least as far as which companies have the means to provide the foundation for generative AI.
“It is absolutely true that few companies will be able to train true frontier models simply because of the resources required,” OpenAI’s Altman told a U.S. Senate committee on Tuesday.
“So I think we and our competitors need an incredibly tight scrutiny,” he added.
