Infrastructure Startups funded by Y Combinator (YC) 2026

Fivetran automates data movement out of, into and across cloud data platforms. We automate the most time-consuming parts of the ELT process from extracts to schema drift handling to transformations, so data engineers can focus on higher-impact projects with total pipeline peace of mind. With 99.9% uptime and self-healing pipelines, Fivetran enables hundreds of leading brands across the globe, including Autodesk, Conagra Brands, JetBlue, Lionsgate, Morgan Stanley, and Ziff Davis, to accelerate data-driven decisions and drive business growth. Fivetran is headquartered in Oakland, California, with offices around the world. 

* Ready to Use Instantly: Skip the complex server setup. It comes pre loaded with top AI Agents (like Claude Code and Hermes) so your agents can start working immediately. * Keep Your Work Safe: Unlike other systems that delete everything when they finish a task, these sandboxes remember your files, code, and settings forever until you delete them. * Pay Only for What You Use: Stop paying for idle servers. You pay a tiny fee to store your files, but you only pay for compute power during the exact minutes your AI is actively running. * Zero Infrastructure Headaches: You don’t need to be a senior DevOps engineer or manage complex cloud networks. One simple command handles the entire backend for you. * Connect to Any App: Allow your AI agents to securely log into and interact with everyday tools like Slack, Gmail, and GitHub in just a few clicks.

we build self-healing browser infrastructure that runs millions of agent operations a day without supervision. teams use it for large-scale data collection, agentic workflows, continuous monitoring, and anything else that requires acting on the web at volume.

Mirrors provides regression testing for AI agents. Teams shipping agents constantly change prompts, models, and tools, and there's no way to test any of it. Agents call bespoke internal tools and APIs no test framework can reproduce, so most teams ship on vibes. Mirrors rebuilds the tools your agent calls from your own production traces, then replays real sessions against every change before it merges. You see which tool calls changed, which arguments moved, and where state landed differently. And when something fails, we tell you whether your agent regressed or something upstream moved. Built for teams whose agents take irreversible actions. Refunds, record updates, outbound email. Runs in your CI on every PR.

Hopper post-trains STT, TTS, and speech LLMs on your production calls, then serves and improves them against your live traffic

Maritime is cloud infrastructure for AI agents. We help companies give every customer an agent running in its own isolated, persistent computer, starting at $1 per agent per month. Maritime handles hosting, scaling, and sleep/wake automatically, so developers can run thousands of agents without managing the infrastructure.

**Problem** Humans used to need laptops to do their work. Now they prompt agents instead, and those agents need computers too: always-on, in large quantities, and with unlimited capabilities. Building and maintaining that yourself is far from trivial. **Solution** Boat provides powerful, simple, permissive Ubuntu VMs in the cloud. In a sandbox, an agent can work for long periods on code or on knowledge work like legal, research, reporting, and auditing. **Why Boat and not any other cloud VM?** 1. It is ready out of the box: every sandbox ships with a desktop, chrome, languages, docker, snapshots, and templates, plus a CLI, SDK, API, and control pane for your fleet. 2. It is permissive and paradigm agnostic: any number of agents can use any number of sandboxes. Agents can run inside or use sandboxes as an external tool. 3. It is expertly crafted: we manage everything from the individual CPU to the orchestration software, which makes sandboxes stable, persistent, scalable, and secure. 4. It is the best performance and price in town: our CPUs rank first on HPC benchmarks, and sandboxes are typically 9x to 18x cheaper than other leading sandboxes. **Further notes** Sandboxes are hosted in Europe, which keeps prices low, and more regions are coming. Unless compliance requires it, agents don't notice the region they run in.

We enable companies monetize their agents' traces.

Familiar is the Neolab for world translation. Dubbing into 30 languages. Now with Familiar, the same actor's performance is preserved instead of a mismatched voice over a mismatched face. One model does the voice, the lips, and the scene together, with identity permanence and scene & world understanding. Built by the researchers who pioneered joint audio-visual human models.

Intelligence will be the most valuable asset a company owns. Today, almost every company rents it. Companies pay a handful of labs one token at a time. Each dollar buys an answer, not an asset. They spend more, but own nothing more. Riften reverses that. Change one environment variable and Riften becomes the gateway for a company's AI traffic. We route each request to the lowest-cost model that can do the job, serving most work on open-weight models we host. Customers save money immediately without rebuilding their products. The gateway is the entry point. Over time, Riften turns the company's work into private models built on open weights and shaped around the business. What begins as cheaper inference becomes an intelligence stack the company owns, runs on infrastructure it controls, and uses to power its most important applications. We call this earned intelligence. It is created by the company's own work, controlled by the company, and accumulated over time. The next generation of companies will not rent their intelligence. They will earn it, own it, and compound it. Riften is how they get there.

Caution is a hosting platform for software you can't afford to get hacked. It simplifies and extends confidential compute, and lets developers deploy workloads to secure enclaves in minutes and give customers, auditors, and counterparties cryptographic proof that production is running the exact reviewed source code, build, configuration, and runtime - not an opaque binary controlled by a cloud operator or internal admin. Confidential computing hardware exists, but the tooling is too fragmented and incomplete for most teams to use well. Today, attestation often proves only that a binary has a certain hash; it does not prove what source code produced that binary. Building the missing deployment, reproducibility, and verification layer requires specialized security engineering and often locks teams into one cloud or hardware stack. Caution turns that into a simple git-based workflow that builds reproducible enclave images, provisions infrastructure, and lets anyone independently verify what is running. We built Caution after years securing high-risk systems through Distrust, where we worked with hedge funds, custodians, blockchain teams, cloud platforms, and critical grid operators on systems responsible for more than $600B in assets. Again and again, the same problem showed up: teams running mission-critical infrastructure could not reliably prove what was executing in production. As AI, fintech, healthcare, and regulated data collaboration move into the cloud, "trust us" is no longer enough. Caution is a general-purpose platform: if it runs on a server, it can run verifiably on Caution. We're starting where trust failures cost the most, but we believe verifiable compute will become the default way software runs, the same way HTTPS became the default way it's served. Infrastructure you can verify, not trust.

OpenVector transforms any existing camera into an automated 24/7 workforce without requiring new hardware. We've built the world's fastest vision-language model inference engine that allows users to simply describe a task in plain English and automatically build a custom AI vision model to monitor their feed. When it detects an important physical event—like a stalled production cycle, a safety violation, or a waiting customer—it operates a computer just like a human to take immediate action. It integrates with legacy systems, makes calls, sends emails, updates a POS/CRM, or triggers webhooks in real time. OpenVector enables any camera to do real work.

hiloop helps teams train agents for tasks where general models are not good enough. Give us a task, your current agent or model, and an evaluation. hiloop runs an autoresearch campaign across data, SFT and other post-training methods, continual learning, prompts, tools, harnesses, and systems, then returns the best verified improvement. It runs hosted or in your cloud. We provide the research system around models: persistent memory, full experiment lineage, compute orchestration, and statistical verification. We’re starting with agent and model training, continual learning, and optimization. Reach out to us for early access at founders@hiloop.ai.

Executor.sh is an MCP gateway that allows you to sign in once to all of your integrations and use them with any client. We enable you to connect your agents to anything (MCP, GraphQL, OpenAPI) so regardless of what you're trying to connect agents to you're able to. If you're trying to scale agents to access all of your production services, data, etc, you need Executor as a unified management layer.

We build GitHub for agents. We're rebuilding the infrastructure for the next hundred billion commits and CI jobs with hundreds of engineers already making the move.

An agent’s access should depend on the task it is doing. The same action can be appropriate for one task and wrong for another. Access decisions need to adapt in real time as the agent works, but companies can’t turn every business judgment into a fixed rules or have people manually approve every step. We train small, company-specific models that approve AI actions in real time. They consider the company’s policies and interests, the agent’s current task, and its past actions to decide whether each proposed action should be allowed. Access is specific to the task at hand and reassessed as the agent works. Companies will rent intelligence from AI labs or run open-source models to power their agents. They’ll own the models that keep their entire AI workforce acting in their interests. Decawork is building those models.

Orchestra is an inference cloud. We capture your AI work traces, train smaller models, and automatically deploy them (only when they outperform the expensive model you are currently paying for). Instead of renting the same frontier model forever, your AI infrastructure gets faster, smarter, and cheaper every time it runs. Suppose your product uses Claude for a repetitive legal workflow. Orchestra observes successful production runs, learns the specific tools and judgment required for that workflow, and trains a specialized open-weight model. Once that model matches or beats Claude on a held-out evaluation, Orchestra starts routing production traffic to it. As more work is completed, the model keeps improving. We started Orchestra because we believe every company using AI should be accumulating intelligence, not accumulating API bills.

Speko is a router for voice models: one API key, and your voice agent runs on the speech-to-text, LLM, and text-to-speech models we measured as best for your language and use case. The measurements are public at benchmarks.speko.ai: we test new models the day they ship and re-route when the numbers change, so you rewrite nothing. We ran this bake-off by hand for four years at previous company, shipping production voice for Fortune 500s.

Inference has surpassed training as AI’s largest compute cost, accelerated by increasing demand for AI and agents. By using proprietary KV cache transfer tech, OneTriangle has cut inference costs by 20% and time by 40% compared to present standards.

For your AI agents to behave like teammates, we give them persistent identities with their own email, phone, and iMessage. We also provide them an agent-to-agent channel, so they can communicate for task delegation.

Edviro builds AI that operates energy infrastructure. We connect to meters, HVAC systems, batteries, solar, and other energy equipment and continuously handle the work required to keep them efficient: detecting and diagnosing problems to forecasting demand, simulating fixes, creating work orders, coordinating maintenance, and controlling equipment where authorized. Our energy world models learn how each site behaves, choose the highest-value actions, and verify the results against real operational, meter, and billing data.

We are building an open source gateway for AI budget holders to control model access and turn their traffic into better models that they own. We're a team that has earned an ML PhD at Georgia Tech, scaled AutoGPT to 160k GitHub stars, and led a self-driving simulation team at Waabi.

Dialogus builds self-improving AI voice agents for enterprise operations. Our agents handle high-volume workflows across customer support, sales, collections, and other operational functions, integrating directly with the systems enterprises already use. Our agents continuously learn from real conversations and business outcomes: identifying where interactions fail, proposing improvements, testing those changes across thousands of simulated conversations, and deploying better versions with human oversight. The result is AI that doesn’t just automate calls, it gets better at doing the job over time.

We train human brain cells for compute

Markov builds expert computer-use datasets for frontier AI labs. Our datasets include screen recordings + mouse/keyboard actions of professionals using CAD, design and enterprise software. We work with the top AI labs and we’ve gotten more than 350k downloads on HuggingFace. Our long-term goal is to be the full stack platform for computer-use AI, starting with data and then the infrastructure to train and deploy computer-use models. Dev and Harish are technical co-founders. Dev studied aerospace engineering at IIT Madras and was the youngest team member at Sarvam - India’s top AI lab. Harish is an Emergent Ventures grantee and his prior robotics work was featured in top national publications.

AI agents are starting to act in the real world, but they can't reliably understand it. Every fact about a place has its own silo: land in GIS maps, power in utility portals, people in census tables, policy in county PDFs. All built for humans, none for agents. Underneath, the data is fragmented across thousands of sources, unstructured and complex. Mireye is building the infrastructure that powers agents to make decisions and act in the real world. One API and MCP to connect agents to the physical world with accurate data enrichment, signals and tools. Our mission is to index every inch of the earth, and make it as queryable as the web.

Buy tokens over a time window from the provider of your choice at a price fixed today - 10-30% off the on-demand price. If your plans change - resell your reserved capacity to another buyer.

machine0 is a CLI for agent compute: `machine0 new mybox` gives your agent a full cloud VM, billed by the minute, from small up to 60 vCPU / 240 GB and 8×H100s. Agents drive it themselves via CLI or MCP. Agent workloads are moving from ephemeral to always-on. A coding agent working on a hard feature runs 6-8 hours. Auto-research agents run for days. machine0 is the environment in which these agents run.

Context.dev gives AI agents structured data from any website through a single API. We help developers build smarter agents & products that depend on accurate, fresh web data. If you need structured data from the internet, you need Context.dev Proud to power agentic products at Mintlify, Super, Vizzy, Klarna, and 600+ other companies.

Miso Labs is building the world’s most emotive foundation models for voice. Our goal is to pass the voice Turing test.

Codag compresses tool call outputs before agents read them, which reduces token spend by 20-30% while retaining output accuracy. Govern and see analytics on agent dollars across all model providers and internal agents.

RightNow AI is a research lab building GPU infrastructure that lets teams own their AI stack instead of depending on closed-source APIs. Our platform RunInfra takes any HuggingFace model, auto-generates optimized GPU kernels, and deploys it serverlessly with pay-per-token pricing RightNow AI lab: https://rightnowai.co RunInfra platform: https://runinfra.ai.

Coding agents in the cloud don't work for majority of teams who require native capabilities; they lack XCode for iOS, Emulator for Android, Unity for game development teams. We are on a mission to cloudify every local development utility so that regardless of where it runs, your agent can compose an ideal set of capabilities on the fly so you don't need to run it on your laptop. Today, we provide remote services such as XCode, iOS & Android simulators to enable coding agents running in any sandbox to be able to build and iterate over mobile apps, too. Replit, Rork, Momentic, Minitap and more agent companies have built mobile development and testing experiences on Limrun platform. Coconote (Quizlet), SorceJobs (W25) and more companies with mobile apps use Limrun with their cloud agents, merging PRs without having to check out the code locally.

Wato gives AI agents shared memory, reusable workflows, connected tools, and living artifacts so work compounds across a team instead of disappearing into chat threads. Teams use Wato to give every agent access to trusted company context and turn useful runs into knowledge the next agent can build on.

The $2T global telecom industry was built for humans. Agents can't act in the real world today because without a phone number, they lack identity. AgentPhone gives every AI agent its own phone number to reach people and businesses through messages and calls.

Expanse unlocks wasted GPU capacity. We recover idle compute through three capabilities: resource prediction (right-sizing job submissions before they reach the scheduler), optimisation suggestions (code and config changes researchers can apply themselves), and failure prediction (catching jobs that will fail before they consume hours of GPU time). We’re four engineers. We ran HPC and GPU training workloads at the largest quant funds and national supercomputing centres. We faced this problem first hand and the only fix was to over-provision and burn millions. Ismaeel built the first multimodal HPC resource predictor as research at EPCC (Edinburgh’s Parallel Computing Centre), which beat every published baseline. This is the tool we wish we had.

We build distributed compute clusters with the cheapest CPUs and GPUs across Hyperscalers and Neoclouds for AI. Our mission is to bring frontier-grade infrastructure to everyone. We're starting by building large scale high performance computing (HPC) clusters for quantitative trading firms to run parallel simulation workloads such as backtesting. Our technology generalizes to critical AI workloads such as post-training with reinforcement learning, fine-tuning, long-horizon agents with high tool use, and batch inference.

Kugel is an enterprise-ready TTS model, available on-prem, with a focus on 30+ languages and low latency.

One memory for your team's agents. Turn every meeting, Claude session, and Slack chat into context your team and agents share.

dreach watches your ATS and your network. when someone starts hiring, you get the manager who owns the role and your warmest way in.

Archal provisions sandboxes with the API environments your CI, tests, and evals need. Your coding agent can start the sandbox & environments, connect to your API integrations, run tests, and inspect state changes. Sandboxes and environments are startable within seconds and take milliseconds to reset.

Inth monitors your pull requests, codebase and website to catch privacy issues and turns them into fixes and evidence. Every product change you make exposes you to privacy risk. Customers lose trust. Fines and lawsuits are given. Fundraising and customer deals are broken. New market launches halt. The growth you thought you would have slows because privacy compliance is overwhelming and changes with every product update. Inth is fixing this to help you automate privacy compliance to ship and grow faster. Inth scans your PRs, website and codebase to catch any privacy risk. It then fixes and evidences them against your product, policies and global privacy laws to automate privacy compliance in every region your customers are in. Start a free privacy scan at: inth.com

Clone any postgres DB regardless of size in 6s so agents can test their work

Smol machines solve the problem of "but app works on my machine" by shipping the machine. Smol machine enables devs to ship a lightweight virtual machine with any software, similar to how Electron ships an entire browser with the web app.

Interfaze is an AI model built on a new architecture that merges specialized DNN/CNN models with transformers for tasks that require deterministic output and high consistency like Document intelligence, Object detection, Classification, Audio understanding and more. Interfaze outperforms large lab models like Gemini flash, Claude Sonnet and the GPT-5.4 mini series Architecture paper: https://arxiv.org/abs/2602.04101 Full open benchmarks: https://interfaze.ai/leaderboards

The AI-native observability platform for fast-moving engineering teams. Backed by tastemakers from the world's top AI companies: Vercel, Graphite, Daytona, Browserbase, LangChain, Mastra, Replit, and more.

Every computer ever made forces you to do one thing at a time. One cursor. One keyboard. One Hardware. One app in focus. Fifty years. Nobody questioned it. We did. We started over. We built InfinityOS, a fundamentally new computer OS on the cloud from ground up. Infinity runs every app in its own independent computer -- own GPU, own input, own environment. Windows and Linux in the same session. Humans and agents working side by side. No concurrency ceiling. Cold start in seconds. runs on the browser of any device..

Hub is a multimodal lab providing real world data to frontier labs and robotic companies.

Building the communication infrastructure needed for fully autonomous agents