The explosion of generative AI has taken the world by storm, but one question is rarely asked: Who can afford it?
OpenAI bled about $540 million last year while developing ChatGPT and says it needs $100 billion to meet its ambitions, according to industry media outlet The Information.
We will be the most capital-intensive startup in the world Silicon Valley History, OpenAI founder Sam Altman told a panel recently.
And when Microsoft, which has poured billions of dollars into investing in OpenAI, is asked how much its venture into AI has cost, the company answers with assurances that it is watching its bottom line.
build something close to the scope of OpenAI, Microsoft or Google On show, it takes an incredible investment in cutting-edge chips and recruiting award-winning researchers.
“People don’t realize that doing a lot of AI stuff like ChatGPT requires a huge amount of processing power. Training these models can cost tens of millions of dollars,” Jack said. goldan independent analyst.
How many companies can actually buy 10,000 companies nvidia H100 systems that go for tens of thousands of dollars a piece? “
The answer is pretty much nobody and in technology, if you can’t build the infrastructure, you rent it and that’s what companies are already doing at scale by outsourcing their computing needs to Microsoft, Google and Amazon’s AWS.
And with the advent of generative artificial intelligence, this reliance on Cloud computing Experts have warned that the tech giants are going too deep, leaving the same players in the driving seat.
Unexpected costs of cloud computing are “a hugely underestimated problem for many companies,” said Stefan Sieg, chief product officer at Software AG, which develops software for businesses.
Sigg compares cloud costs to electricity bills and says companies that don’t know better will be in for a “big surprise” if they allow their engineers to run up bills in the mad rush to build technology, including artificial intelligence.
Microsoft’s signature cloud offering is Azure and some observers believe the giant’s all-out bet on AI is really about protecting the success of Azure and ensuring the future of the cash cow.
Azure has been the unexciting winner for the baking giant for years, turning in huge profits but without grabbing iPhone or iPad titles Social media that go directly to the consumer.
For Microsoft, “The golden goose is monetizing the cloud with Azure because we’re talking about what could be a $20, $30, $40 billion opportunity a year in the future if the AI bet works,” said Dan Ives of Wedbush Securities.
Microsoft CEO Satya Nadella insists that generative AI is “moving quickly in the right direction.”
Ives predicted that Nadella, who is well respected on Wall Street, would have a grace period of six or nine months to prove his bet a winner.
Microsoft acknowledges the risks, but insists that as far as artificial intelligence is concerned, it must “lead the wave,” Chief Financial Officer Amy Hood told analysts this month.
“We’re going to charge for these AI capabilities, and then eventually, we’re going to generate an operating profit,” she said.
Accumulating profits in the company founded by Bill Gates can only mean passing the cost of artificial intelligence to customers.
From Main Street to the Fortune 500, relying on AI-amped will be expensive, and companies and investors are working on alternatives to at least reduce the bill.
“AI training, GPT training will become a very important cloud service in the future,” said Tenri Fu, CEO of Spectro Cloud.
His company, like many others in the sector, helps companies optimize cloud technology to reduce expenses.
“But after training, the company will be able to restore its model to implement real AI,” he added, hoping that dependence on cloud giants will be reduced.
The regulators hope that they can continue, and not leave the giants in charge, imposing their terms on the smaller ones.
“Enforcers must ensure … that opportunities and opportunities for competition … are not trampled upon by incumbents,” FTC Chair Lena Khan told CNBC.
But it may be too late, at least when it comes to companies that have the means to provide the foundation for generative AI.
“It is absolutely true that the number of companies that can train real frontier models will only be small because of the resources required,” OpenAI’s Altmann told a US Senate committee on Tuesday.
“And so I think there has to be incredible scrutiny on us and our competitors.”