CEO Announces $102.5M Series A Funding for Together AI

CEO Vipul Ved Prakash Announces $102.5M Series A Funding for Together AI

CEO Vipul Ved Prakash, co-founder of Together AI, proudly announces the closure of a remarkable $102.5 million Series A funding round. This substantial investment is set to power the company’s vision of an open-source future for AI.

“This funding marks a significant milestone for Together AI,“ states Vipul Ved Prakash. “It empowers us to accelerate our mission of creating an open and comprehensive cloud platform for developers worldwide.“

Kleiner Perkins spearheaded this round, with key contributions from prominent investors like NVIDIA and Emergence Capital. Notably, Bucky Moore from Kleiner Perkins joins the board, amplifying the company’s strategic direction and guidance.

Bucky Moore, expressing excitement about AI’s impact on software development, remarks, “Together AI is poised to facilitate the adoption of open-source models. This will empower organizations to build powerful applications on this innovative foundation.“

Since its launch in June 2023, Together AI has witnessed exponential adoption of its AI products by startups and enterprises. With this infusion of capital, the company aims to expedite the development of the fastest cloud platform for generative AI applications.

The funding round saw a coalition of investors, including NEA, Prosperity 7, Greycroft, 137 Ventures, and Seed investors such as Lux Capital, Definition Capital, Long Journey Ventures, SCB10x, SV Angel, Factory, and Scott Banister, underlining the widespread interest and confidence in Together AI’s trajectory.

Bucky Moore emphasizes the transformative nature of AI in software development, stressing the need for universal accessibility. He anticipates open-source models gaining traction as their performance rivals closed alternatives, positioning Together AI as a catalyst for organizations building applications on these models.

Today’s landscape sees a rising demand among startups and enterprises for a generative AI strategy free from vendor lock-in. With its platform enabling seamless integration of open source models or custom model creation, Together AI empowers developers to own and integrate their AI investments effortlessly.

Joseph Floyd of Emergence Capital emphasizes, “Together AI is well-positioned for enterprises investing in generative AI. The platform offers control over proprietary IP while scaling AI investments.“ He draws parallels between past shifts to cloud applications and the current surge in generative AI investments.

Fueling Together AI’s industry-leading performance is their steadfast focus on core research. The release of the RedPajama-V2 dataset, a substantial open dataset for training LLMs, has garnered immense interest within the AI community.

Chief Scientist Tri Dao‘s work on FlashAttention v2, adopted by industry giants like OpenAI, Anthropic, Meta, and Mistral, further solidifies Together AI’s position in advancing leading LLMs. Their innovative inference techniques have propelled the platform to offer the fastest inference stack for transformer models.

In tandem with research, Together AI places immense emphasis on robust compute infrastructure, operating at 20 exaflops across multiple US and EU data centers. Their custom-designed cloud infrastructure, powered by NVIDIA GPUs and AI cloud partners, ensures optimal performance for high-complexity AI applications.

Looking ahead, Together AI envisions generative AI as a transformative platform technology with far-reaching implications for human society. Emphasizing the coexistence of proprietary and open models, the company remains dedicated to enabling global researchers and developers to shape the AI future actively. “Our mission is to enable researchers and developers worldwide to actively shape the AI future,“ concludes Vipul Ved Prakash.

Insights from Industry Observers and Experts

According to industry observers, Together AI’s strategy revolves around prioritizing training within the AI infrastructure sector. They have highlighted that “99% of current AI infrastructure spend is in training (vs inference),” suggesting a significant opportunity for the company.

Discussions also emphasized the crowded nature of the inference provider space, noting that “the big cloud providers are very well positioned to dominate” due to minimal entry barriers. However, experts note the unique nature of Together AI’s approach, citing the complexities of training as “a different ball game,” requiring specialized expertise and infrastructure.

Moreover, discussions around the imbalance in compute spend between training and inference prompted questions about the reasons behind it, and one query posed was, “Why? Is it that improvements are so frequent it makes sense to toss out older versions?”

NewsletterYour weekly roundup of the best stories on AI. Delivered to your inbox weekly.