Baseten's $1.5B raise challenges OpenAI's grip with 20x revenue growth

For SaaS and cloud providers, Baseten is rewriting the AI infrastructure playbook. The startup's $1.5 billion funding round — its fourth in 18 months — and a $13 billion valuation reflect a seismic demand for AI inference that doesn't rely on the high costs of OpenAI or Anthropic. As enterprises race to embed custom models into their workflows, Baseten's promise of cheaper, scalable infrastructure is disrupting the unit economics that have long favored the big AI labs. This is the kind of pricing leverage that could reshape cloud AI budgets and redefine platform loyalty.

Baseten, a California-based AI infrastructure startup co-founded by Australians, has closed a staggering $1.5 billion funding round that values the company at $13 billion. The round, led by U.S. investment firms Sands Capital and Wellington Management, marks the biggest single investment ever by Australian venture capital firm Blackbird VC. The deal underscores the frenzied capital pouring into AI, particularly into the inference layer where trained models are deployed in real-world applications. Baseten's specialty is providing software and infrastructure that allows enterprises to customize and run their own AI models at a lower cost than using heavyweights like OpenAI and Anthropic. The company reported that its revenue has grown 20-fold over the past year, a figure that directly fueled the sky-high valuation. This is Baseten's fourth funding round in just 18 months, a breakneck pace that signals intense investor conviction — and intense pressure to scale.

“

The startup's $1.5 billion funding round — its fourth in 18 months — and a $13 billion valuation reflect a seismic demand for AI inference that doesn't rely on the high costs of OpenAI or Anthropic.

The broader context is a seismic shift in the AI market. While the initial AI boom focused on training large language models, the industry is now pivoting to inference, the stage where models produce outputs for actual business tasks. Inference is expected to account for the vast majority of long-term compute costs, and companies that can make it cheaper and more accessible stand to capture enormous value. Baseten is positioning itself as the cost-effective alternative to closed-model providers, offering unit economics that undercut those of OpenAI and Anthropic. This is not just a pricing battle; it's a fight over the future architecture of enterprise AI. By allowing companies to deploy custom models without the expensive APIs of the major labs, Baseten could fragment the market and redistribute value toward infrastructure firms.

Blackbird partner Michael Tolo described the investment as 'a signal of conviction' and called the competitive leverage Baseten provides 'the biggest shift that we've seen in both unit economics and competitive leverage within the AI market so far.' The statement is revealing: it comes not from a Silicon Valley insider but from Australia's most prominent VC, signaling that the AI gold rush has fully globalized. Blackbird's decision to make its largest-ever bet outside its home market, on a company that operates in the infrastructure layer, is a vote of confidence that the picks-and-shovels play is far from over.