Sam Altman’s ambitious target of securing approximately $7 trillion for the development of artificial intelligence (AI) chips highlights the extreme challenges and high stakes in the AI industry.
The introduction of ChatGPT in late 2022 ignited fierce competition and led to a surge in new companies entering the generative AI space.
However, it’s anticipated that many of these newcomers will either cease operations or merge with established entities soon.
Consider the example of Sasha Haco, CEO of Unitary, a company that monitors videos on social platforms for content violations.
The expense for Unitary to utilize OpenAI‘s video analysis AI tools would be a hundredfold higher than its client fees. Consequently, Unitary has opted to develop its models, which requires a delicate balance of resources and capabilities.
The company relies on leasing specialized AI chips through cloud services provided by giants like Microsoft Corp. and Amazon’s AWS, which have seen their prices double since 2020 and are often hard to secure.
Haco revealed that there have been instances where access to necessary resources was so limited that costs surged tenfold.
While Unitary manages to sustain its operations, Haco acknowledges that no generative AI startup has yet mastered the art of scaling its business cost-effectively, unlike major tech corporations.
Another AI entrepreneur from San Francisco shared that profitability for some startups dependent on rented AI chips and cloud services hinges on minimal usage of their products by consumers.
Ronald Ashri, CEO of Dialogue.ai, which specializes in custom chatbots for regulated sectors, likened the situation to electricity consumption.
He explained that the constant use of foundational models, which act as a source of ‘electricity’ for their services, represents the largest expense in delivering solutions to their clients.
