Sam Altman
’s aspiration to amass approximately $7 trillion for the production of
AI chips
unveils a narrative that extends beyond his seemingly audacious goals. As a report in Bloomberg says, firstly, the infrastructure required for AI development has become prohibitively costly. Secondly, a significant portion of this value is still reportedly predominantly controlled by a few major tech
companies
, and this oligopoly is set to intensify.
Despite the competitive environment ignited by the introduction of ChatGPT in late 2022, and the surge of new startups entering the much-hyped generative AI market, the report says that it is likely that most of these newcomers will either collapse or be absorbed by the existing players in the coming year or so.
The big likely reason behind the struggle
Reason being: As Bloomberg's report says that the operational costs are simply too steep for them to sustain independently.
It cites example of Sasha Haco, the CEO of Unitary, a company that monitors social media videos for violations. The cost for her company to use OpenAI’s video-scanning AI tools is said to be be 100 times more than what they charge their clients. Consequently, Unitary creates its own models, which is a precarious endeavor in itself. Her startup needs to lease those scarce AI chips from cloud providers like Microsoft and Amazon.com Inc's Amazon Web Services. According to Haco, these chips have seen a twofold price increase since 2020 and are hard to secure.
While Unitary manages to operate, Haco told the publication that no generative AI startup has yet cracked the code on how to run a cost-effective business at scale, at least not in the way that large tech firms have.
It's win-win for Microsoft, Amazon, Nvidia and Google
Generative AI startups are reported to have two options for building their technology. They can either create their own version of OpenAI’s GPT-4 or Google’s Gemini, known as a foundation model, which requires an investment of hundreds of millions of dollars. Alternatively, they can build on an existing model, which requires an investment of tens of millions and is the route most AI startups currently take.
In both scenarios, the main beneficiaries are cloud-computing behemoths Microsoft, Amazon, and Alphabet Inc.'s Google, along with AI chip manufacturer Nvidia Corp. “At present, all these startups are taking money from venture capital investors and handing it over to cloud companies and Nvidia,” told Rodolfo Rosini, CEO of chip company Vaire Computing. This is why Nvidia’s shares have more than doubled in the past year, bringing it close to a $2 trillion valuation.
English (US) ·