“Otherwise, all the incumbents will win,” Mr. Conrad said.
Venture capital investors have a similar goal. This month, Index Ventures teamed up with Oracle to provide a mix of Nvidia’s H100 chips and an older version, called the A100, to its very young portfolio companies for free.
Erin Price Wright, an investor at Index Ventures, said the firm has seen its start-ups struggle to navigate the complex process of acquiring computing power and landing on waiting lists as long as of nine months. Two companies are slated to use the company’s new program, with others expressing interest.
Before the shortage, George Sivulka, chief executive of Hebbia, an AI productivity software maker, simply asked his cloud provider for more “instances,” or virtual servers loaded with GPUs, as the company expanded. Now, he said, his contacts at cloud companies have not responded to his requests or added him to a four-month waiting list. He resorted to using customers and other connections to help make his case to cloud companies. And he keeps looking for more.
“It’s almost like talking about drugs: ‘I know a guy with H100s,'” he said.
A few months ago, some Hebbia engineers set up a server with some less efficient GPUs in the company’s Manhattan office, parked the machine in a closet and used it to work on smaller projects . Liquid cooling units keep the server from overheating, Mr. Sivulka said, but it’s noisy.
“We closed the door,” he said. “Nobody sits next to it.”
Scarcity creates a stark contrast between the haves and the have-nots. In June, Inflection AI, an AI start-up in Palo Alto, Calif., announced that it had acquired 22,000 of Nvidia’s H100 chips. It also said it had raised $1.3 billion from Microsoft, Nvidia and others. Mustafa Suleyman, chief executive of Inflection, said in an interview that the company planned to spend at least 95 percent of the funds on GPUs.