Generative AI has become increasingly popular in recent years, with companies like OpenAI, Microsoft, and Google investing heavily in its development. However, the question of who can afford to invest in such technology is rarely discussed. OpenAI reportedly lost around $540 million last year while developing ChatGPT and estimates that it will require $100 billion to achieve its goals. This makes it one of the most capital-intensive start-ups in Silicon Valley history, according to its founder, Sam Altman.
Building something similar to what OpenAI, Microsoft, or Google have developed requires a significant investment in state-of-the-art chips and hiring award-winning researchers. This can cost tens of millions of dollars, making it difficult for most companies to afford. For instance, 10,000 Nvidia H100 systems cost tens of thousands of dollars per unit, which is beyond the budget of most organizations. As a result, many companies outsource their computing needs to cloud computing providers like Microsoft, Google, and Amazon’s AWS.
The Impact of Cloud Computing
The advent of generative AI has increased companies’ reliance on cloud computing providers, leaving them in the driver’s seat. However, the unpredictable costs of cloud computing are a problem for many companies, according to Stefan Sigg, Chief Product Officer at Software AG. Such costs are comparable to electricity bills and can come as a big surprise to companies that let their engineers run up bills in the rush to build tech, including AI.
Microsoft’s Azure is a signature cloud offer, and some observers believe that the company’s all-in bet on AI is about protecting Azure’s success and guaranteeing the cash cow’s future. Azure has been Microsoft’s unsexy breadwinner for years, generating huge profits without attracting the headlines of an iPhone or social media. For Microsoft, the golden goose is monetizing cloud with Azure, which could generate $20, $30, or $40 billion annually down the road if the AI bet is successful, according to Dan Ives of Wedbush Securities.
Microsoft CEO Satya Nadella believes that generative AI is moving fast in the right direction and has a six- or nine-month grace period to show that his bet is successful. While Microsoft acknowledges the risk, it must lead this wave in AI, according to CFO Amy Hood. The company will charge for those AI capabilities and ultimately deliver operating profit, which means passing on the cost of AI to customers.
From Main Street to Fortune 500, companies’ dependency on AI will be expensive, and they are exploring alternatives to reduce the bill. Spectro Cloud CEO Tenry Fu suggests that AI training, GPT training, will become a vital cloud service going forward. Companies can optimize cloud technology to reduce expenses, but after training, they can get their model back for real AI application, hopefully reducing their dependence on cloud giants.
Regulators hope to keep up with the giants and not leave them in charge, imposing their terms on smaller companies. FTC chairwoman Lina Khan believes that law enforcers must ensure that opportunities and openings for competition are not squashed out by the incumbents. However, it might be too late, at least when it comes to which companies have the means to provide the groundwork of generative AI. Altman told a US Senate panel that the number of companies that can train the true frontier models will be small because of the resources required. Therefore, there needs to be incredible scrutiny on OpenAI and its competitors.
While generative AI has taken the world by storm, it is an expensive technology that requires significant investment in state-of-the-art chips and hiring award-winning researchers. Most companies cannot afford to invest in such technology, leading them to outsource their computing needs to cloud computing providers. However, the unpredictable costs of cloud computing are a problem for many companies, and regulators must ensure that opportunities and openings for competition are not squashed out by incumbents.
Leave a Reply