The explosion of productive AI has taken the world by storm, but one question rarely arises: Who can afford it? OpenAI spent about $540 million last year to develop ChatGPT and says it needs $100 billion to achieve its goals, The Information reports.

“We will be the most capital-intensive venture in Silicon Valley history,” OpenAI founder Sam Altman said at a panel recently.

And when asked how much its AI adventure will cost Microsoft, which has invested billions in OpenAI, the company says it’s looking at the bottom line. Building anything even close to the scale of OpenAI, Microsoft or Google will require a huge investment in cutting-edge chips and the involvement of award-winning researchers.

“People don’t realize that building a significant amount of AI like ChatGPT takes a lot of computing power. “And training these models can cost tens of millions of dollars,” said Jack Gold, an independent analyst.

“How many companies can buy 10,000 Nvidia H100 systems costing tens of thousands of dollars each?” Gold asked.

The answer is almost nobody, and in technology, if you can’t set up the infrastructure, you lease it, which companies already do in bulk, subcontracting computing needs to Microsoft, Google, and Amazon AWS. Experts say that with the advent of productive artificial intelligence, this dependence on cloud computing and technology giants has deepened and left the leadership to the same players.

“Seriously underrated”

The unforeseen costs of cloud computing “are a largely underestimated problem for many companies,” said Stefan Sigg, chief product officer of Software AG, which makes business software. Sigg compares cloud costs to electricity bills and says companies that don’t know better will be in for a “big surprise” if they let their engineers collect the bills in a frenzied rush to build technologies, including artificial intelligence.

Microsoft’s signature cloud offering is Azure, and some observers believe the AI ​​giant’s mention is actually aimed at preserving Azure’s success and providing a future cash cow. Azure has been an unattractive breadwinner for years, reaping huge profits but capturing iPhone or direct-to-consumer social media headlines.

“The golden goose is monetizing the cloud with Azure because we’re talking about potentially $20,30,40 billion per year going forward if the AI ​​bet is successful,” said Dan Ives of Wedbush Securities for Microsoft. Microsoft CEO Satya Nadella insists that productive AI is “moving rapidly in the right direction”.

Ives predicted that Wall Street-respected Nadella would have a grace period of six to nine months for his bet to show that he had won. Microsoft acknowledges the risk but insists AI must “lead this wave”, CFO Amy Hood told analysts this month.

“We will charge for these AI capabilities and eventually drive operating profits,” he said.

Saving profits in the company that Bill Gates founded may only mean passing on the value of artificial intelligence to customers. AI-powered addiction from Main Street to the Fortune 500 will be expensive, and companies and investors are looking for alternatives to at least lower the bill.

“Artificial intelligence training, GPT training will become a very important cloud service in the future,” said Tenri Fu, CEO of Spectro Cloud.

Like many companies in the industry, his company helps companies optimize their cloud technologies to reduce costs.

But after learning, the company hopes it will be able to bring its model back to real-world AI applications and reduce reliance on cloud giants.

Regulators hope they can keep up with the giants by imposing their terms on small companies.

“Implementers … must ensure that competitive opportunities and opportunities … are not blocked by incumbents,” FTC Chairman Lina Hahn told CNBC.

But it may be too late, at least when it comes to which companies have the tools to lay the foundation for productive AI.

“It’s certainly true that few companies will be able to train truly advanced models just because of the resources involved,” OpenAI’s Altman said at a US Senate panel Tuesday.

“So I think we and our competitors need to do an incredible amount of due diligence,” he added.