Y Combinator’s New defense technology — Request for Startups (Summer 2024) was a focused invitation for founders building products relevant to defence, national security, and dual-use markets. Although the original Summer 2024 batch has passed, the request remains useful as a blueprint for understanding how a top accelerator evaluates technically ambitious defence startups in 2026.
For Indian founders, the opportunity is broader than applying to one historical call. The same principles apply when approaching accelerators, strategic investors, defence primes, government programmes, and procurement channels in India and overseas.
What the request was looking for
The request encouraged startups working on difficult problems where software, artificial intelligence, hardware, and operational expertise intersect. Relevant areas can include:
- Autonomous systems for sensing, navigation, inspection, logistics, or mission support.
- Cybersecurity and cyber defence, including detection, response, identity, and infrastructure resilience.
- Defence manufacturing and advanced hardware, such as robotics, sensors, communications equipment, and materials.
- Geospatial intelligence, remote sensing, computer vision, and decision-support systems.
- Simulation and training for operators, analysts, and emergency-response teams.
- Dual-use infrastructure that serves both government and commercial customers.
The strongest applications are not simply “AI for defence.” They identify a specific operational failure, explain who experiences it, and show why the proposed product can work in real conditions.
Why Indian founders should care in 2026
India’s defence ecosystem is becoming more accessible to startups through innovation challenges, public procurement reforms, defence corridors, research partnerships, and a growing network of private manufacturers. However, defence sales still involve long validation cycles, compliance obligations, field testing, and demanding integration requirements.
A startup should therefore avoid treating an accelerator application as a generic fundraising exercise. It should use the application to demonstrate that the team understands:
- The end user and the environment in which the product will operate.
- The difference between a prototype, a tested system, and a deployable capability.
- Data ownership, security classification, export controls, and responsible-use constraints.
- How the product integrates with existing command, communications, or operational systems.
- Which commercial or government buyer can approve and pay for deployment.
A dual-use approach can improve resilience. For example, a perception system might first serve industrial inspection, disaster response, or logistics before expanding into defence. Founders building an AI-heavy product should also plan a reliable production stack; this 2026 guide to AI startup tech stacks covers practical choices around infrastructure, deployment, and cost.
How to assess your startup’s fit
Use a simple five-part test before applying to an accelerator or approaching a defence customer.
1. Is the problem operationally important?
Describe the problem in terms of time lost, mission risk, personnel exposure, false alarms, downtime, or procurement cost. “Improves situational awareness” is weak. “Cuts the time required to identify and classify an incident from 30 minutes to five” is stronger.
2. Does the product work outside a demo?
Defence environments may involve poor connectivity, limited compute, harsh weather, adversarial inputs, and incomplete data. Explain test conditions, failure rates, latency, battery life, uptime, and human override. If your system uses language or vision models, document how it behaves when confidence is low.
3. Is there a credible distribution path?
Identify whether the first customer is a government department, prime contractor, security operator, manufacturer, or commercial enterprise. Name the pilot structure, procurement route, integration partner, and expected sales timeline without overstating certainty.
4. Does the team have unusual insight?
Y Combinator and defence investors look for founders with technical depth and direct exposure to the problem. Relevant experience may come from engineering, military service, aerospace, cybersecurity, public safety, industrial operations, or research. A small team can be compelling if it can build and validate quickly.
5. Can the business expand beyond one contract?
A product dependent on a single bespoke deployment may be difficult to scale. Explain the reusable platform, recurring software component, manufacturing plan, or adjacent market that supports growth.
What to include in the application
A clear application should answer the following questions directly:
- What does the company build, and for whom?
- What is the most important evidence that it works?
- Why is the problem urgent now?
- Why can this team solve it better than incumbents?
- What has been built, tested, sold, or deployed?
- How will the company reach its first ten customers?
- What regulatory, security, safety, or integration barriers remain?
- What will the company accomplish in the next six to twelve months?
Use concrete metrics rather than broad claims. Include a short product video or field-test evidence when appropriate. If confidentiality limits what you can disclose, state the limitation and provide verifiable alternatives such as anonymised results, customer references, or controlled demonstrations.
For AI products, explain the data pipeline, evaluation method, model constraints, and human review process. A defence buyer will care about reliability and auditability more than a benchmark score. Teams building edge systems should also plan for low-bandwidth operation, model updates, logging, and secure rollback.
Funding and programme expectations
Do not rely on outdated claims about a specific batch’s terms. Y Combinator’s programme structure and funding terms can change, so applicants should confirm current details on the official programme page before submitting. More importantly, treat accelerator capital as support for product development and validation—not as a substitute for procurement readiness.
Budget realistically for hardware iterations, testing, certifications, security reviews, insurance, manufacturing tooling, and travel to customer sites. Government revenue may take longer than expected, so maintain a disciplined runway and consider commercial pilots where they do not compromise the core mission.
A practical 30-day preparation plan
- Days 1–7: Define the operational problem, target user, buyer, and measurable outcome.
- Days 8–14: Build or refine a working prototype and run tests under realistic constraints.
- Days 15–21: Collect customer interviews, pilot evidence, technical metrics, and security requirements.
- Days 22–26: Write a concise application, record a product demonstration, and pressure-test the go-to-market plan.
- Days 27–30: Ask technical and domain experts to challenge your assumptions, then submit only after removing unsupported claims.
If your product automates intelligence, reporting, or internal operations, study approaches to AI workflow automation for high-growth startups. If voice or field communications are central to the product, the principles in this guide to cost-effective voice AI can help you evaluate latency, infrastructure, and operating cost.
Final takeaway
The Summer 2024 request should be read as a signal about the kind of defence technology accelerators want: technically ambitious, operationally grounded, and capable of becoming a large business. In 2026, Indian founders should pair a strong prototype with evidence of field performance, a realistic procurement strategy, and a clear responsible-use framework. The best application will make the product easy to understand, difficult to dismiss, and credible beyond the accelerator interview.