Meta is rolling out a new initiative aimed at encouraging startups to adopt its cutting-edge Llama AI models. Dubbed Llama for Startups, this program offers companies direct support from Meta’s Llama team along with financial assistance in certain cases. Eligible U.S.-based firms, which are incorporated, have raised less than $10 million in funding, employ at least one developer, and are focused on developing generative AI applications, can apply by the May 30 deadline.
According to a blog post by Meta, selected members may receive up to $6,000 per month for a maximum of six months to help cover the costs associated with developing and enhancing their generative AI solutions. Meta’s experts will collaborate closely with these startups to kickstart their projects and explore advanced use cases for Llama that could benefit their businesses.
Despite Meta’s success in amassing over a billion downloads of its Llama models, the company faces stiff competition from the likes of DeepSeek, Google, and Alibaba’s Qwen in the open model space. Recent setbacks, including delays in releasing the flagship AI model Llama 4 Behemoth and allegations of cheating in an AI benchmark, have further complicated Meta’s efforts to solidify its position in the market.
Looking ahead, Meta has ambitious revenue targets for its generative AI products, forecasting earnings of $2 billion to $3 billion by 2025 and a staggering $460 billion to $1.4 trillion by 2035. The company has also launched revenue-sharing agreements with hosting partners for its Llama models, introduced an API for customizing Llama releases, and hinted at potential monetization strategies for its AI assistant powered by Llama, including displaying ads and offering subscription services with enhanced features, as revealed by CEO Mark Zuckerberg during a recent earnings call.
The development and maintenance of these innovative products come at a significant cost for Meta. In 2024, the company’s “GenAI” budget exceeded $900 million, a figure that is expected to surpass $1 billion this year. These expenses do not even include the infrastructure required to operate and train the AI models, with Meta projecting capital expenditures of $60 billion to $80 billion in 2025, primarily allocated to new data centers.
