top of page
Lanon Wee

OpenAI Backed by Nvidia, Eric Schmidt Could Emerge as Winner After Closed Door Drama

Mistral AI, meanwhile, launched in late 2019 and secured $3 million in a seed round in December, led by Salesforce Ventures. Investment from tech bigwigs like Salesforce, Qualcomm, Nvidia and Eric Schmidt, the former Google CEO, is being funneled into open source Artificial Intelligence (AI) startups such as Hugging Face and Mistral AI. The OpenAI situation has been resolved with Sam Altman returning to the role of CEO and the establishment of a new board, but it has stirred up the market and has caused companies to reassess the wisdom of depending on one, closed-source service when it comes to utilizing generative AI as a competitive weapon. Bring in IBM and Amazon Web Services (AWS) as customers of Hugging Face and you have a platform which can claim over 2 million users, 400,000 models and 77,000 data sets. Mistral AI, launched in late 2019, recently received $3 million in a seed round, under the sponsorship of Salesforce Ventures. The startup, which spun out of the MILA research lab in Montreal, is led by Joelle Pineau, professor of computer science at McGill University, who also heads the lab. Tech giants Salesforce, Qualcomm, Nvidia, and high-profile investor Eric Schmidt have made sizable investments in open source AI startups, which could gain even more ground following the OpenAI saga. Now with Sam Altman back as CEO and a board being formed with Microsoft's approval, the market is seeing a shift in the reliance on single, proprietary services for generative AI, as well as questions about the power of a handful of big tech players in AI development. As competition increases and more enterprises consider open models and providers for AI, many of these well-funded open source startups are looking to expand. Mike Gualtieri, vice president and principal analyst at Forrester, explains the importance of open source or community models that are outside of a single company. This drama with OpenAI demonstrates the fragility of relying on a single entity for AI needs, a sentiment echoed by AI research scientist and academic Delip Rao. He emphasizes the importance of disruption-proof technology that only open source can provide. Big bets on open source AI have been around for a while and include Meta Platforms' Llama, an open source AI model controlled by one dominant tech company. Salesforce-invested Hugging Face is gearing up after raising $400 million from 30 tech investors, including Qualcomm, IBM, Google, Nvidia, Intel, and Sequoia Capital. Thomas Wolf, co-founder and CEO, says that they are not pursuing the Silicon Valley playbook and are playing for the long run. He also emphasizes that they are not trying to control and own software models, rather they want to unlock the community to develop AI. In addition to Salesforce Ventures's $500 million generative AI fund, former Google CEO Eric Schmidt is backing Mistral AI, another open source rival to OpenAI's ChatGPT. Founded by ex-Alphabet Deep Mind and Meta alums, the startup quickly raised $113 million from Lightspeed Venture Partners and other Silicon Valley investors. Microsoft has also offered the service on its Azure cloud computing platform. Poolside AI, a startup that spun out of the MILA research lab in Montreal, recently raised $126 million and is headed by Joelle Pineau, professor of computer science at McGill University. The OpenAI saga has brought to light the dangers of relying on a single entity for AI needs. Instead, open source models and providers outside of one company should be encouraged in order to create a more resilient and long-term AI ecosystem. In France -- which is initiating a government-business partnership to become a major global AI player -- Hugging Face, Mistral and Poolside had three of Europe's largest AI startup funding deals this year. To further this aim, former Google president Kai-Fu Lee launched 01.AI, funded by Sinovation Ventures and Alibaba, this November in China; and OpenAI rival xAI, backed by Elon Musk, is being tested with a limited number of users. Hugging Face, which started out as a chatbot, shifted gears to use NLP models, and has since developed its open source platform for machine learning and data science. More than 2 million users, 400,000 models and 77,000 data sets have been registered on the platform. Most of Hugging Face's operations are free, though there are some that bring in revenue (undisclosed). Companies like IBM and AWS make use of the tool to integrate into their own offerings. Quinn Li, senior vice president and global head at Qualcomm Ventures, was quoted as saying, "Collaborative efforts can increase the speed of development and app-based AI." Wolf, CEO of Hugging Face, commented that AI is revolutionizing the world, and "Gen AI" is covering the whole range of AI. In relation to open source AI models, venture capitalist Bill Gurley, along with a number of investors and innovators, sent a letter to Joe Biden this month that drew attention to the possibilities of the recent AI Executrive Order being used to restrict open source. They argued that such a model of development is essential for promoting competition, democratizing usage of the technology, and encouraging creativity. Revolution CEO Steve Case added that with open source software, developers are able to collaborate and share information, leading to a speedy process of innovation. At the same time, while open source models can be regarded as transparent and trustworthy, they are also more exposed to bad actors. As such, the challenge for startups that work with giant tech firms is whether they can stay impartial when contributing to open source. Professor Thomas Malone noted that in tech-related developments, it is hard for prominent figures to stay prominent, and the tussle between the advantages of markets and the power of the community is a factor that must be taken into account.

Comments


bottom of page