Amazon has announced that it will invest up to $230 million in startups developing generative AI-powered applications.
About $80 million of the investment will go toward Amazon's second AWS Generative AI Accelerator program, aimed at positioning AWS as an attractive cloud infrastructure option for startups developing generative AI models to power products, apps, and services. The majority of the new funding (including the full amount set aside for the accelerator program) will be provided in the form of AWS infrastructure compute credits, which cannot be transferred to other cloud service providers such as Google Cloud or Microsoft Azure.
Additionally, Amazon has committed to providing startups participating in this year's Generative AI Accelerator cohort access to experts and technology from Nvidia, the program's presenting partner. Companies will also be invited to participate in the Nvidia Inception program, where they will have the opportunity to connect with potential investors and additional consulting resources.
The Generative AI Accelerator program has also seen significant growth: Last year's cohort of 10 startups received up to $300,000 in AWS compute credits, representing a total investment of nearly $3 million.
“With this new effort, we are giving startups the building blocks they need to help launch and scale world-class businesses and enable new AI applications that impact every aspect of how the world learns, connects, and does business,” Matt Wood, vice president of AI products at AWS, said in a statement.
Amazon's increased spending on generative AI technologies, including initiatives such as the $100 million AWS Generative AI Innovation Center and free credits for startups using its flagship AI model and Project Olympus, comes as the company tries to catch up with larger tech rivals in the growing and increasingly competitive field of generative AI. Amazon claims its various generative AI businesses have achieved “multi-billion dollar” profit margins, but the field is widely perceived as a missed opportunity.
According to The Information, AWS had originally planned to unveil its own generative AI model, similar to OpenAI's ChatGPT (codenamed Bedrock) – which eventually became Amazon's Bedrock model hosting service – at its annual conference in November 2022. However, a critical bug forced the company to postpone the launch.
Amazon's Alexa division has also been struggling with technical glitches and political infighting, Fortune's Sharon Goldman reported this week. Nine months after a splashy press demo of the “next-generation” Alexa, the new Alexa is reportedly still nowhere near primetime due to a lack of training data, lack of access to training hardware, and other obstacles.
Amazon also turned down early investment opportunities in two big AI startups, Cohere and Anthropic, and the company later tried to invest in Cohere but was turned down, forcing it to settle for a co-investment (albeit a massive $4 billion total) with arch-rival Google.
Besides the recent departure of Howard Wright, AWS' head of startups, who managed the company's relationships with startups, another obstacle for Amazon is growing regulatory scrutiny of big tech companies' investments in AI startups.
The US Federal Trade Commission recently opened an investigation into Microsoft's backing of OpenAI and Google and Amazon's investment in Anthropik, and European policymakers have also expressed skepticism about the deals.