California’s Applied Compute, a company that assists firms in developing, deploying, and improving open-weight AI models, is reportedly negotiating with investors to secure financing valued at about $3 billion, according to a report from The Information.
If raised, this would more than double its former valuation of $1.3 billion in a funding round disclosed only four months ago, demonstrating changing perceptions by investors concerning AI infrastructure.
This is an indication of a wider change in the AI industry. With the advancement of open models, businesses are constantly searching for systems that they can personalize for their use and manage, resulting in a growing demand for infrastructure that can train, deploy, and improve those models.
Hugging Face, the leading platform for open models, stated that in 2025 its platform hit 13 million users and has over 2 million public models as proof of the advancement of the open-source AI ecosystem.
Doubling up from a $1.3 billion Series B
Applied Compute reported an $80 million funding round in April, with Kleiner Perkins being the major financier at a post-money valuation of $1.3 billion, which raised the company’s overall funding to $160 million.
In the new round, the $3 billion figure values the company at approximately 2.3x that level within months. The funding is still open, so this valuation could change.
The proposed figure is also in stark contrast to the funding data from Epoch AI. Mistral AI, one of the leading developers of open models, raised $3 billion in equity as of September 2025, at a valuation of $13.7 billion according to the Epoch dataset.
In contrast, Applied Compute has so far raised about $160 million, which means that its estimated valuation of $3 billion is based on much less funding when compared to Mistral. The companies themselves cannot be compared directly, as Mistral creates foundation models while Applied Compute gives the necessary infrastructure, yet such a difference does show the level of importance that investors pay to the infrastructure for open-source AI models.
Mistral’s most recent round of financing was almost 25 times more than the total funding received by Applied Compute and had a valuation roughly 4.6 times more than Applied Compute’s. This gives an understanding of the valuation increase of Applied Compute without having to rely on uncertain revenue projections.
What Applied Compute actually sells
Applied Compute describes its platform as a cloud for training, inference, and continuous improvement of open models. Its Agent Cloud, or AC2, lets enterprises train models around their own data and workflows while deploying them in production.
Its main idea lies in linking model training to the deployment of agents. In contrast to the traditional practice of differentiating training from deployment, Applied Compute allows organizations to keep their agent infrastructure unchanged and deploy an alternative trained model that fits their needs.
“The harness can stay where it already runs.” — Vinjai Vale, Applied Compute
The platform uses production traces, enterprise context and reinforcement learning to improve specialized models. That could become increasingly valuable as businesses move beyond generic AI assistants.
Applied Compute’s research illustrates the approach. In a recent experiment, it trained a router to distribute software-engineering tasks among NVIDIA’s Nemotron 3 Ultra, GPT-5.5 and Claude Opus 4.7. The company said the router achieved GPT-5.5-level performance at roughly 25% lower cost while capturing most of the benefit of an oracle strategy.
The strategy treats AI models as interchangeable components. Instead of sending every task to the most powerful model, enterprises can balance capability and cost.
That becomes more relevant as open models narrow the gap with proprietary systems. Cryptopolitan has previously covered Kimi K3’s competition with OpenAI and Anthropic, showing how Chinese open-weight models are increasingly challenging US systems on capability and price.
Why the timing favors an open-model bet
Applied Compute’s fundraising discussions come as AI infrastructure spending shifts toward production workloads. Gartner expects inference to account for 55% of AI-optimized infrastructure-as-a-service spending in 2026 as AI moves into more real-world applications.
That trend fits Applied Compute’s model. The company is not simply selling training compute; it is building infrastructure designed to keep models improving after deployment.
Hugging Face’s growth offers another indication of the expanding market. Its platform reached 13 million users and more than 2 million public models in 2025, as developers increasingly build fine-tuned models and applications on existing systems.
For enterprises, open-weight models also offer greater control. Companies can customize them around proprietary workflows and deploy them while retaining more control over their data and model behavior.
That gives Applied Compute a potentially valuable position. If businesses increasingly want to build AI around their own data rather than simply rent intelligence from closed providers, infrastructure connecting models to proprietary workloads could become an important part of the global AI market.
The proposed $3 billion valuation is therefore more than a bet on one startup. It is a bet that companies will increasingly use, train, customize and continuously improve AI models around their own businesses.
Be the first to comment