Why generative AI orchestration startups are poised for growth in 2024
No Content
Startups are racing to bring generative AI to other companies. Plenty have stumbled along the way.
Many of the most pressing challenges have less to do with perfecting large language models than sorting out how to use them. Choosing the right model, integrating multiple LLMs into a single service, and deploying applications at low cost—these are the problems projects now face.
Enter orchestration startups, which help developers integrate and manage generative AI models. And they’re poised to play an ever-larger role as founders embrace a multi-model approach and more companies put generative AI into action.
“Open-source generative AI orchestration projects will create multiple unicorn startups in early-stage deals,” PitchBook senior analyst Brendan Burke predicts in PitchBook’s 2024 Enterprise Technology Outlook. “Given how much future value has been assigned to model developers, expectations for leading orchestration frameworks should become similarly outsized.”
Orchestration startups run the gamut from model fine-tuning, application development and deployment, and hardware acceleration. As more startups look to cut down on costs by opting for cheaper open-source models than OpenAI’s ChatGPT or Anthropic’s Claude, orchestration startups are allowing users to access multiple models and route tasks and queries wherever they choose. Since proprietary models often charge by usage, many startups are utilizing routing as a way to cut down on costs.
One such orchestration startup, Anyscale, has raised a total of roughly $260 million, according to PitchBook data, with backing from Andreessen Horowitz, NEA and Intel Capital. Anyscale’s platform is used by OpenAI and others for the development, training and fine-tuning of AI. The startup raised a $199 million Series C at a $1.01 billion valuation in 2022, according to PitchBook data.
Ion Stoica, co-founder and executive chair of Anyscale, says that orchestration startups are set to grow as generative AI advances.
“There is this continuously growing gap between the demands of these [LLM] applications and the abilities of hardware,” said Stoica, who also co-founded Databricks and is a professor at the University of California, Berkeley. “LLMs have now far surpassed compute demands.”
Stoica says that orchestration platforms narrow that gap by providing several services in one. Instead of having a team of data scientists, orchestration platforms like Anyscale allow anyone to tweak, input and develop when it comes to large language models.
Other notable rounds for orchestration startups include Weights & Biases’ $50 million raise in August, Pinecone’s $100 million round in April, and Essential AI’s $57 million funding announced in December.
“Our pitch,” Stoica says, “is solving a fundamental problem which is going to be more important tomorrow than it is today.”
This post was originally published on Why generative AI orchestration startups are poised for growth in 2024 | PitchBook
