TL;DR:
- Together, a startup focused on open-source generative AI solutions, raised $20 million in funding.
- The startup aims to develop open-source generative AI models and services to integrate AI into production applications.
- Together plans to build a cloud platform for executing, training, and fine-tuning open-source models with scalable computing resources at lower costs.
- The co-founders emphasize the importance of open and decentralized alternatives in generative AI to address privacy, transparency, customization, and ease of deployment.
- Together’s projects include RedPajama, GPT-JT, and OpenChatKit, which focus on developing open-source generative models and chat models.
- The company aims to bring greater transparency, control, and privacy to AI and believes that collaboration and open ecosystems are crucial for AI’s advancement.
- Together faces competition from other open-source models and startups offering AI hardware infrastructure.
- The company plans to expand its team and allocate funding for research, development, infrastructure, and product enhancement.
- Together’s solution aims to meet the requirements of large organizations and generate interest from enterprises seeking transparency, control, and privacy in AI.
Main AI News:
Generative AI has become a magnet for investors, drawing significant attention and funding. Recent data reveals that generative AI startups secured an impressive $1.7 billion in funding during the first quarter of 2023. Moreover, there are several players in the market, including established companies like OpenAI and Anthropic. Despite the fierce competition, venture capitalists remain enthusiastic about supporting emerging and untested contenders.
A notable example is Together, a startup dedicated to developing open-source generative AI solutions. In a recent announcement, Together revealed that it successfully raised $20 million in a seed funding round. This substantial investment was led by Lux Capital, with participation from other prominent firms such as Factory, SV Angel, First Round Capital, Long Journey Ventures, Robot Ventures, Definition Capital, Susa Ventures, Cadenza Ventures, and SCB 10x. Additionally, notable angel investors, including Scott Banister, co-founder of PayPal, and Jeff Hammerbacher, a founding employee of Cloudera, also contributed to the funding round.
Lux Capital’s Brandon Reeves shared his perspective on Together’s impact, stating, “Together is spearheading AI’s ‘Linux moment’ by providing an open ecosystem across compute and best-in-class foundation models.” Reeves further emphasized that the Together team is fully committed to fostering a vibrant open ecosystem that allows individuals and enterprises alike to participate actively in the advancement of AI.
Founded in June 2022, Together is the brainchild of Vipul Ved Prakash, Ce Zhang, Chris Re, and Percy Liang. Prakash, the former senior director at Apple after his company Topsy was acquired by the tech giant in 2013, brings valuable experience in social media search platforms.
Zhang, an associate professor of computer science at ETH Zurich, is currently on sabbatical and leading research on “decentralized” AI. Re, a serial entrepreneur, has co-founded multiple startups, including SambaNova, which specializes in hardware and integrated systems for AI. Lastly, Liang, a computer science professor at Stanford University, serves as the director of the university’s Center for Research on Foundation Models (CRFM).
Together aims to facilitate the integration of AI into production applications by creating open-source generative AI models and services. The startup is building a cloud platform that enables the execution, training, and fine-tuning of open source models. The co-founders claim that their platform will offer scalable computing resources at significantly lower costs compared to dominant vendors such as Google Cloud, AWS, and Azure.
Vipul Ved Prakash emphasized the importance of open and decentralized alternatives in the generative AI field, stating, “We believe that generative models are a consequential technology for society, and open and decentralized alternatives to closed systems are going to be critical to enable the best outcomes for AI and society.” Prakash further explained that enterprises seeking to adopt generative AI strategies require privacy, transparency, customization, and ease of deployment, which are not adequately addressed by existing cloud offerings with closed-source models and data.
This sentiment aligns with the growing realization among incumbents that open source AI initiatives pose a significant challenge. An internal memo leaked from Google suggests that both Google and its competitors are unable to compete with the long-term potential of open source AI. Furthermore, OpenAI is reportedly preparing to publicly unveil its first open source text-generating AI model, reflecting the increasing popularity of open source alternatives in the market.
Together’s first major project, RedPajama, aims to cultivate a collection of open source generative models, specifically focusing on “chat” models similar to OpenAI’s renowned ChatGPT. Collaborating with esteemed institutions such as the MILA Québec AI Institute, CRFM, and ETH’s data science lab, DS3Lab, RedPajama commenced by releasing a dataset that enables organizations to pretrain models with permissive licensing.
Another notable endeavor by Together is GPT-JT, a modified version of the open source text-generating model GPT-J-6B, initially released by the research group EleutherAI. Additionally, Together embarked on OpenChatKit, an ambitious endeavor to develop an equivalent to OpenAI’s ChatGPT.
Vipul Ved Prakash, the visionary behind Together, highlights the challenges associated with training, fine-tuning, and commercializing open source generative models. He emphasizes that the current solutions demand significant expertise in AI and the ability to manage extensive infrastructure. However, Together’s platform aims to alleviate these challenges by providing an accessible and user-friendly solution out-of-the-box.
While the efficacy and seamlessness of Together’s platform are yet to be seen as it has not officially launched, some argue that the company’s efforts might appear somewhat redundant in the larger AI landscape. The number of open source models available from community groups and prominent laboratories grows rapidly, albeit not all of them are licensed for commercial use. Nevertheless, various alternatives, like Databricks’ Dolly 2.0, exist for organizations seeking open source AI models.
In terms of AI hardware infrastructure, startups such as CoreWeave claim to offer high-performance computing at competitive rates, challenging the dominance of major public cloud providers. Furthermore, community-driven free services for running AI text-generating models have emerged. Taking inspiration from these community groups Together plans to develop the Together Decentralized Cloud, a platform that aims to pool hardware resources, including GPUs contributed by volunteers across the internet.
So, what sets Together apart? Vipul Ved Prakash argues that the platform brings greater transparency, control, and privacy to the table. This value proposition echoes that of startup Stability AI, which directs computing power and capital toward open source research while commercializing and offering services built on top of these products.
Prakash explains that regulated enterprises will be significant customers of open source models, as they enable organizations to thoroughly inspect, comprehend, and customize models to suit their specific applications. He firmly believes that overcoming the challenges in AI can only be achieved through a global collaborative effort. Consequently, Together has made it its mission to establish and nurture a self-sustaining, open ecosystem that generates the most advanced AI systems for the betterment of humanity.
Undoubtedly, Together’s goals are ambitious, and the company is still in its early stages, refraining from disclosing any current customers or revenue. Nevertheless, the company is pressing forward, with plans to expand its team from 24 to approximately 40 employees by the year’s end. The seed capital will primarily be allocated to research and development, infrastructure enhancement, and product development.
Vipul Ved Prakash affirms that Together’s solution, built on open source generative models, is meticulously crafted to meet the requirements of large organizations, providing them with a robust foundation for their generative AI strategies. He further adds that enterprises are displaying tremendous interest in Together, seeking greater transparency, control, and privacy in their AI endeavors.
Conlcusion:
The success and significant funding raised by Together, along with the growing interest in open-source generative AI, underscore the immense potential and opportunities in the market. Together’s focus on providing accessible and cost-effective solutions for integrating AI into production applications aligns with the increasing demand for transparency, control, and privacy.
As more enterprises seek to adopt generative AI strategies, the emergence of open-source alternatives and decentralized ecosystems will likely reshape the market dynamics, challenging incumbents and fostering a collaborative environment. This signifies a shift towards a more democratized AI landscape, where innovation and customization are key drivers for organizations aiming to leverage the power of generative AI.
