Blaxel raises $7.3M in a seed round for ‘AWS for AI Agents’ after processing billions agent requests

Blaxel (19459046) has already shown significant traction. By the end of Y Combinator, they had processed millions of agent requests per day across 16 global regions. One customer runs over 1 billion seconds for each agent to process millions videos. This scale illustrates the infrastructure requirements of AI-first businesses.

“One customer processes session replays so that product managers can better understand the user behavior for their product,” Sinai explained. “They must process millions of replays each month.” It’s millions of minutes. They are using our agentic architecture to process these session replays and provide insight for product managers.”

This company’s strategy focuses on providing infrastructure that AI Agents can operate themselves rather than requiring administrators. This includes sandboxed, virtual machines that boot up in less than 25 milliseconds. It also includes automatic scaling based upon agent activity patterns and APIs designed for AI systems to consume directly, rather than human developers.

How six cofounders with an exit plan can take on Big Tech.

Blaxel’s unusual six-founder model is a result of the team’s shared experience in building and selling their previous company. OVHcloud Europe’s largest cloud provider. This company became OVH Analytics, allowing the team to gain first-hand experience of both cloud infrastructure challenges as well as successful exits.

I know it sounds strange, but we are a pretty large team. Sinai, referring to Y Combinator’s signature event, said that the team didn’t quite fit on the stage during demo day. “But we did that.” “My previous company, which i sold to OVH Cloud, we also six cofounders.”

This team includes Charles Drappier who Sinai has known for more than 15 years, as well as co-founders Christophe Ploujoux Nicolas Lecomte Thomas Crochet and Mathis Joffre. Their collective experience spans across infrastructure, developer tools, platform engineering, and more — crucial expertise for competing with tech giants who have virtually unlimited resources. Sinai said, “I think that it’s important that we are six at this time because we have a great deal of ambition.”

“We are building the next generation of cloud computing to meet this new agentic age.”

What makes Blaxel stand out in the competitive cloud-infrastructure market

? The cloud infrastructure industry is fiercely competitive. AWS holds roughly a third of the market, while newer players such as Modal, Replicate and RunPod are focusing on AI workloads. Blaxel differentiates by focusing on AI agents instead of model inference or learning.

“Most competitors you mentioned are tackling a very hard problem, which is how to host your model and how to make those models as quickly as possible in terms of the number of tokens,” Sinai explained. “But not many people are working on infrastructure for agents, which is exactly what we do.” Serverless APIs MCP (Model Context Protocol), serversto connect agents with external tools and a unified gatewayfor accessing multiple AI models. The infrastructure is designed for AI agents to have variable resource requirements. They may require minimal computing power when waiting for responses, but significant resources while actively processing.

Enterprise security and compliance features are targeted at regulated industries.

Blaxel, despite targeting AI-first startups, has implemented enterprise-grade measures such as SOC2 and HIPAA. The platform provides data residency controls, which allow customers to restrict workloads within specific geographic regions. This is critical for companies in regulated industry. Sinai explained that “we provide a policy framework which you can attach to workloads in order to say this agent cannot be run outside of these subsets of region.” You can attach a rule that says this agent cannot run outside the United States. This will ensure that the agent only processes data in the regions that you specify. Sinai said that it is important for young companies to have the best infrastructure and practices because they will become enterprises.

Pay-as you-go pricing offers 50% cost savings compared to traditional serverless

Blaxel adopted a pay as you go pricing model similar with established cloud providers. They shifted away from a subscription-based approach after validating the market demand during their Y Combinator batches. The model only charges customers when their agents are actively completing tasks. Infrastructure is shut down during idle periods in order to optimize costs.

We provide infrastructure that can be spun up in a few milliseconds, and shut down within a second,” Sinai said. “So, you only pay for the time that your agent is actually processing a request.” You don’t pay when your agent is waiting on something else because we shut it off.

This approach has already saved customers money. One client achieved a 50% cost reduction in comparison to typical serverless solution while processing terabytes data monthly.

Gartner predicts that 75% of apps will be using AI agents by 2028.

Industry analysts predict explosive growth in AI agent usage. Gartner predicts that By 2028, 75% of applications will be developed with AI agents though Sinai believes that current enterprise adoption is still largely experimental. He said that most companies currently working in production are smaller companies and not yet enterprise companies. “So, we are focusing on serving them just like the big cloud provider did in the past.” Amazon Web Services focused initially on startups and developer friendly companies before expanding into enterprise customers. Blaxel will follow a similar route, using the $7.3million to expand their software before possibly moving into custom hardware or data center optimization. Sinai said that seven million dollars is not enough money to build a data center, but it’s important to take things step-by-step. “We want to make sure we have the best possible interfaces for our customers and the best services for their agent, before we go into deeper infrastructure optimization.” Blaxel believes that with the projections of hundreds and billions of AI agents over the next decades, it is possible to build infrastructure from the ground-up for this new computing paradigm.

“We think that there is an enormous economy that is starting around agents,” Sinai said. “There will be hundreds of billions AI agents and the infrastructure that we have today is not designed for this new phase.”

Y Combinator, Liquid2 , Transposeand angel investors who are in line with the company’s vision for a purpose-built agent infrastructure. Blaxel’s specialized approach may allow it to capture a significant market share as AI agents transition from being experimental tools to production systems that handle critical business processes.