The competition for generative AI is getting more aggressive, and it’s not just the prominent tech corporations interested in staking a claim. There’s a lot of money to be made in this field, and it’s coming from unexpected sources.
Recently, the open-source AI platform ‘Hugging Face’ announced a partnership with AWS. They have pledged to “make AI Open and not the other way around.” The collaboration will enable Hugging Face to merge their models with AWS’s cloud, allowing developers to utilize any of the models available on Hugging Face via the AWS-managed ML service, ‘SageMaker.’
The day after, OpenAI launched a new developer platform allowing users to run the latest models, such as GPT-3.5 and a few Da Vinci variants on specific capacity. The startup’s service is known as ‘Foundry,’ Although they have not disclosed their cloud provider, it’s probably Microsoft Azure.
Although both OpenAI and Hugging Face’s services seem to focus on democratizing AI, there is a significant difference. OpenAI, which Microsoft supports, has developed Foundry for “leading-edge customers running big businesses.
Furthermore, this computing power won’t be affordable. A leaked pricing sheet for Foundry surfaced on Twitter, revealing that even the lightweight version of GPT-3.5 costs $78,000 for three months or $264,000 for one year.
Foundry has already attracted its first customers, with both the social media messaging platform Snapchat and Coca-Cola planning to use the service to integrate GPT-3.5 into their business operations.
Hugging Face and AWS have different viewpoints on making generative AI models available to everyone. Hugging Face aims to democratize generative AI and make it accessible to individuals and businesses alike. AWS, on the other hand, believes that generative AI has the potential to transform industries but is costly and requires expertise that only select companies possess.
In a joint statement, Adam Selipsky, CEO of AWS, acknowledged the potential of generative AI but also highlighted its high cost and technical requirements, making it inaccessible to most companies.
Hugging Face’s CEO, Clement Delangue, agreed with this sentiment and stressed the importance of accessibility and transparency in sharing progress and creating tools that can utilize these new capabilities wisely and responsibly. Delangue said SageMaker and AWS chips will help Hugging Face and the larger machine-learning community create reproducible models anyone can build on.
The company is taking bigger steps in democratizing the AI field and recently introduced an API for users to build ChatGPT technology for their applications and products. The API is priced at USD 0.002 per 1,000 tokens or 750 words, making it a ten times cheaper model than their GPT-3.5 variants and accessible to tiny startups and researchers.
ML engineer Mark Tenenholtz praised OpenAI’s efforts and tweeted, “I’d give my left eye to read a book written on the challenges of deploying GPT-3 and the engineering endeavor it took to drop the price 10x with ChatGPT’s API.”
Although the API model had access only to certain model weights, gradients, and training data, it was still a boon for companies without ML experts. However, as OpenAI fights to maintain supremacy in generative AI, Hugging Face is exploring other avenues to keep the AI research space open.
Today, the EleutherAI group of researchers announced the formation of a nonprofit foundation named the ‘EleutherAI Institute.’ The decentralized group has received support from independent companies, including Hugging Face and Emad Mostaque’s Stability AI, former GitHub CEO Nat Friedman, Lambda Labs, and Canva. Despite this backing, these companies remain independent in their respective business segments.
To refer to Hugging Face as a mere success would be an understatement. In May of last year, the company raised a staggering USD 100 million in its Series C round, which resulted in a valuation of USD 2 billion.
Partnering with prominent companies such as AWS is advantageous to Hugging Face in two ways: it generates additional revenue and prevents potential monopolies. In addition to the earnings generated by its subscription model, the company’s collaboration with AWS keeps them ahead of the competition.