The race to build the infrastructure that actually runs AI models has produced one of the year's largest funding rounds. Baseten, a startup whose platform serves and scales open-source and custom AI models, has raised $1.5 billion in a Series F that values the company at up to $13 billion, according to multiple reports published June 22 and 23, 2026.

Reuters reported the deal on June 23, with Bloomberg, The Wall Street Journal, and Tech in Asia corroborating the headline figures. The round underscores how investor attention — and capital — is shifting from the labs that train models to the companies that make those models run quickly and cheaply in production. For more context on this story, see our ongoing latest AI developments.

Why Inference Is the New Battleground

If training a model is the headline-grabbing spectacle, inference — the act of actually running a model to generate a response — is where the money is spent every single day. Every query that flows through a chatbot, a coding assistant, or an AI-powered application triggers inference compute. As adoption has surged, inference has quietly become the largest recurring cost in the AI stack.

Baseten's own tagline captures the thesis plainly: "Inference is everything." The company's Baseten Inference Stack promises the fastest model runtimes, cross-cloud high availability, and seamless developer workflows, underpinned by custom kernels, advanced decoding techniques, and caching built directly into the platform.

The Wall Street Journal, which reported on the company earlier in June, characterized Baseten as betting on cheaper alternatives to OpenAI's hosted models — essentially arguing that organizations will increasingly run capable open-source models on optimized infrastructure rather than paying premium rates to a single provider.

A Record Bet From Down Under

One of the most striking elements of the round is the involvement of Blackbird, the Australian venture capital firm. Reuters reported that Blackbird's participation represents a record bet — believed to be the largest single investment ever made by an Australian VC fund. The commitment signals that the appetite for AI infrastructure deals has spread well beyond Silicon Valley, with international funds willing to write enormous checks for companies positioned at the center of the inference economy.

The Series F propels Baseten into the upper ranks of privately held AI companies. A $13 billion valuation places it alongside some of the most richly valued infrastructure and tooling startups in the sector, and the $1.5 billion in fresh capital gives it the resources to compete on compute, talent, and engineering at a time when the inference market is consolidating rapidly.

What Baseten Actually Does

According to the company's website, Baseten provides an end-to-end platform for deploying AI models in production. Its offerings span several layers:

  • Inference deployment: Organizations can test workloads, prototype products, and evaluate the latest AI models optimized for production speed.
  • Training and one-click deployment: Users can train models and deploy them on inference-optimized infrastructure in a single step.
  • Monetization APIs: Model developers can stand up an inference API powered by Baseten to commercialize their work faster.

The company emphasizes that achieving the fastest inference takes more than just acquiring GPUs. It requires custom kernels tuned to specific models, the latest decoding algorithms, and sophisticated caching — capabilities that Baseten has folded into its platform to differentiate itself from raw cloud compute providers.

The Broader Inference Boom

Baseten's mega-round is the latest sign that the inference layer of the AI stack has become a fiercely contested market. As the cost of running models has emerged as a top concern for enterprises — prompting companies to rein in developer usage and seek efficiency gains — a crop of startups has emerged promising faster, cheaper, more reliable model serving.

The competitive dynamics are intense. Cloud giants including Amazon, Google, and Microsoft offer their own inference services, while specialized players compete on price, latency, and developer experience. Baseten's bet is that a purpose-built inference stack, optimized across the entire pipeline from kernel to API, can outperform generalist offerings — and that open-source models will continue eroding the dominance of any single closed provider.

Whether that thesis pays off will depend on execution, but the market is clearly sizing the opportunity in the tens of billions. With its Series F, Baseten has secured both the capital and the validation — from investors spanning three continents — to make its case that inference, not training, is where the next great AI company will be built.

---

Stay Ahead of AI

Get the latest AI news, analysis, and breakthroughs — all in one place.

Read more AI news →