Generative AI — AI that can write essays, create artwork and music, and more — continues to capture the attention of major investors. Generative AI startups raised $1.7 billion in the first quarter of 2023, with an additional $10.68 billion in deals, according to a source announced in the quarter but not yet completed.
There is a lot of competition, including incumbents like OpenAI and Anthropic. But despite that, VCs don’t shy away from untested players and newcomers.
case in point, Together, a startup developing open source generative AI, announced today that it has raised $20 million – on the larger side for a seed round – led by Lux Capital with participation from Factory, SV Angel, First Round Capital, Long Journey Ventures, Robot Ventures, Definition Capital, Susa Ventures, Cadenza Ventures and SCB 10x. Several high-profile angel investors were also involved, including Scott Banister, one of the co-founders of PayPal, and Jeff Hammerbacher, a co-founder of Cloudera.
“Together, leading AI’s ‘Linux moment’ by providing an open ecosystem for computing power and the best base models,” Lux Capital’s Brandon Reeves told TechCrunch via email. “Together team is committed to creating a vibrant open ecosystem in which everyone from individuals to companies can participate.”
Launched in June 2022, Together is the brainchild of Vipul Ved Prakash, Ce Zhang, Chris Re and Percy Liang. Prakash previously founded the social media search platform Topsy, which was acquired by Apple in 2013, where he later became a senior director. Zhang is an associate professor of computer science at ETH Zurich, currently on sabbatical and leading research in “decentralized” AI. As for Re, he co-founded several startups, including SambaNova, which builds hardware and integrated systems for AI. And Liang, a Stanford computer science professor, directs the university’s Center for Research on Foundation Models (CRFM).
With Together, Prakash, Zhang, Re and Liang aim to create open source generative AI models and services that, in their words, “help organizations integrate AI into their manufacturing applications.” To that end, Together is building a cloud platform to run, train, and refine open source models that the co-founders claim will provide scalable computing power at “dramatically lower” prices than the dominant vendors (e.g., Google Cloud, AWS, Azure).
“We believe that generative models are a consistent technology for society and that open and decentralized alternatives to closed systems will be critical to enabling the best outcomes for AI and society,” Prakash told TechCrunch in an email interview. . “As enterprises define their generative AI strategies, they are looking for privacy, transparency, customization and ease of implementation. The current cloud offering, with closed-source models and data, does not meet their requirements.”
He has a point, at least as far as incumbents are feeling the pressure. An internal Google memo leaked earlier in the month implies that the search giant – and its rivals for that matter – can’t compete with open source AI initiatives in the long run. Meanwhile, OpenAI is reportedly preparing to publicly debut its first open source text-generating AI model amid a proliferation of open source alternatives.
One of Together’s first projects, Red pajamas, aims to advance a range of open source generative models, including “chat” models along the lines of OpenAI’s ChatGPT. A collaboration between Together and several groups, including the MILA Québec AI Institute, CRFM, and ETH’s data science lab, DS3Lab, RedPajama began the release of a dataset that enables organizations to pre-train models that can be licensed permissively.
Together’s other efforts to date include GPT-JT, a fork of the open source text-generating model GPT-J-6B (released by the research group EleutherAI), and OpenChatKit, an attempt at a ChatGPT equivalent.
“Today, training, refining or producing open source generative models is a huge challenge,” said Prakash. “Current solutions require you to have a lot of expertise in AI and at the same time be able to manage the necessary large-scale infrastructure. The Together platform addresses both challenges out-of-the-box, with a user-friendly and accessible solution.”
However, how seamless Together is remains to be seen – the platform has yet to launch in GA. And, you could say, the efforts are a little bit double in the context of the wider AI landscape. The number of open source models from both community groups and major labs growing day by day, practical. And while not all of them are licensed for commercial use, there are several, such as Dolly 2.0 from Databricks.
In terms of AI hardware infrastructure, startups like CoreWeave claim to offer powerful computing power for below-market rates alongside the major public cloud providers. There have even been attempts to build community-driven, free services for running AI text-generating models. (Together plans to follow in the footsteps of these community groups by building a platform tentatively called the Together Decentralized Cloud that will pool hardware resources, including GPUs from Internet volunteers.)
So what brings Together to the table? More transparency, control and privacy, says Prakash. It’s a sales pitch not unlike the sales pitch of startup Stability AI, which funnels computing power and capital into open source research while commercializing and selling the various finished products.
“Regulated enterprises will be big customers of open source, as open source models pre-trained on open datasets allow organizations to fully inspect, understand and adapt the models to their own applications,” he said. “We believe that AI challenges can only be overcome by a global community working together. That is why we have made it our mission to build and manage a self-sustaining, open ecosystem that produces the best AI systems for humanity.”
It’s a lofty goal, to be sure. And it’s early days for Together, which wouldn’t say if it currently has customers — much less revenue. But the company is making progress and plans to increase the size of the team from 24 to about 40 by the end of the year, with the rest of the seed money spent on R&D, infrastructure and product development.
“The Together solution, based on open source generative models, is built on understanding the requirements of large organizations and addressing each of those needs to provide enterprises with the core platform for their generative AI strategy,” said Prakash. “Together is seeing tremendous interest from companies seeking greater transparency, control and privacy.”