There’s no denying the fascination with the success stories of tech startups starting in garages and rising to become global behemoths. From Hewlett-Packard to Google, these tales of humble beginnings have fueled the dreams of many budding entrepreneurs for years.
However, the extensive capital and computational resources required by startups seeking to innovate in today’s hottest tech area – artificial intelligence (AI), as used in chatbots like ChatGPT and Google Bard – might be putting an end to these inspiring narratives.
In 2019, Aidan Gomez and Nick Frosst departed from Google to set up Cohere, an AI startup intended to rival their previous employer. A few months down the line, they found themselves returning to Google to request the massive computational power necessary for the development of their own AI technology.
The agreement was green-lighted by Google’s CEO, Sundar Pichai, enabling them to access the required resources. This illustrates how challenging it is to create a pioneering AI startup without the backing of “hyperscalers”, entities that command expansive data centers capable of powering AI systems.
This situation reaffirms the dominant position of big tech companies, just as the industry is about to undergo a monumental transformation. These companies, boasting thousands of specialized computer chips, are vital for enhancing and broadening the capabilities of technologies like ChatGPT.
As the race to match or outdo GPT-4 intensifies, the question of who will triumph is being hotly debated among entrepreneurs, investors, and industry analysts. While OpenAI is widely regarded as the frontrunner, a handful of smaller companies are also making strides in this area.
Given their vast resources, the tech behemoths are poised to drive these systems to unprecedented levels, reaffirming their stronghold in the AI domain.