Most founders apply to a16z Speedrun like it is Y Combinator with a shorter timeline. That is the wrong frame, and it is why most of them get auto-filtered in the first review pass. At under 1 in 250 odds and up to $1M on the table, the gap between a strong application and a filtered one is not effort - it is frame.

Speedrun is demo-first. It rewards runnable ideas with evidence of velocity, not narrative. The written form is just the qualifier; your demo and your two-minute intro video are what actually carry the application. Get that one distinction right and everything about how you apply changes.

This is the playbook - the real terms, what the partners select for, what the twelve weeks look like, and exactly how to build the video that does the heavy lifting.

What a16z Speedrun actually is

Speedrun started as a16z's bet on games and interactive media and has become its flagship program for AI-native startups. It is run by the a16z Games and Speedrun team, and it is deliberately built for founders shipping fast in the AI era - not polished, later-stage companies.

Key takeaway: the terms are founder-friendly (no board seat, real follow-on capital, huge credits), but the funnel is brutal. Roughly 1 in 250 applicants gets in. You do not win that by being thorough on the form. You win it by being undeniable on the demo.

The thesis: Speedrun is demo-first

A traditional accelerator application rewards narrative - a clean deck, a tidy market-size slide, a story about the team. Speedrun asks a different question: can this founder ship, and are they shipping fast? Answer it with a working product and you are in a different pile from everyone describing a product that does not exist yet.

The evidence is everywhere in how the program is built. The check rewards velocity: half of it is tied to you raising a next round within 18 months. The program is a sprint, roughly one focused day a week for twelve weeks, not a semester of classes. The final Demo Day is a two-minute pitch to a room of investors. And after the initial review, what they ask strong applicants for is a short video pitch or a 15-minute interview - not more paperwork.

Add it up and the conclusion is simple: the signal that carries your application is not the form. It is a working demo plus a crisp two-minute explanation of what you built and how fast you built it. Everything below is built to help you land in the winning pile.

Where Speedrun over-indexes: the categories it wants

Speedrun grew out of a16z's games and interactive-media practice, and that DNA still shapes what gets in. It is broader than games now, but the center of gravity is consumer-facing and creative AI, not enterprise plumbing. The patterns that show up most:

• Games and interactive entertainment - the original mandate, still strong.
• Consumer AI - apps that put generative capability directly in a user's hands.
• Creative and media tools - anything that helps people make images, video, audio, or worlds faster.
• AI developer tools and infrastructure - the picks-and-shovels layer for AI builders.
• Vertical and B2B AI - AI applied to a specific industry with a real wedge.

Key takeaway: if you are building consumer or creative AI, you are swimming with the current. If you are enterprise-heavy, you can still get in, but your velocity and demo have to be even louder to stand out against the program's natural lean.

What Speedrun actually selects for

Speedrun partner Andrew Chen has been blunt that acceptance is not about resumes or credentials - the strongest founders show one of two things: real traction, or exceptional ability demonstrated through past work and domain expertise. As Chen puts it, "any traction and any validation is incredibly helpful proving your case." Three signals come up again and again, in this order.

1. Velocity and traction (the strongest signal)

More than pedigree, Speedrun wants evidence you move fast. Shipped product, real users, week-over-week growth, revenue if you have it. A live thing people already use beats a beautiful pitch for a thing that does not. If you have any traction - signups, usage, early revenue - lead with it, and show the slope, not just the number.

2. Founder quality and unfair insight

They index on founders with a genuine edge in their domain - deep, specific expertise over brand-name resumes. "I worked at a big company" matters less than "I have spent five years inside this exact problem and know something the market doesn't." Complementary co-founding teams are preferred, but strong solo founders get in - and even recruiting one great co-founder is a signal in itself. As Chen notes, "convincing one smart person to join you is probably step one in validating that it's a serious effort."

3. An AI-native idea that could be huge

Speedrun is looking for companies built for the AI era across games, consumer, B2B, dev tools, and creative software. The bar is not "uses AI" - it is a product whose existence makes sense now because of what AI just made possible, with a path to being very large.

Key takeaway: pedigree is a tiebreaker, not the filter. Velocity is the filter. Everything in your application should answer "how fast are you moving, and what have you already shipped?"

Timeline and how to apply

Speedrun runs two cohorts a year and reviews on a rolling basis - typically within 4-6 weeks of your submission. The most recent cohort, SR007, runs July 27 to October 11, 2026 (its application window closed May 17, 2026). If you missed it, you are applying to the next batch - and because review is rolling, applying early in the window is a real advantage.

The application itself is an online form on the official a16z Speedrun site: founder bios, what you are building, and company details. A product/demo link and a short intro video are optional on paper but, given everything above, they are the highest-leverage part of your submission. After the first pass, strong applicants are asked for a video pitch or a 15-minute interview.

How to build the 2-minute intro video that gets you in

Everything above points to one asset: the short intro video. Speedrun keeps coming back to it - the demo link on the form, the follow-up video pitch, the two-minute Demo Day. A form cannot prove you shipped something real. Ninety seconds of your product working can. So this is not an afterthought to bolt on at the end. It is the application. Here is how to build one that lands.

The 5-beat script that works

Good vs bad, in one line each

Technical founders often stall here, because polishing a video feels like time stolen from building. But in a program that selects for speed, shipping a clean two-minute explainer in an afternoon is a velocity signal - it proves you can package and communicate as fast as you can code. If you would rather not fight a video editor, Flowjam turns a raw screen recording into a launch-ready intro video fast, so the demo carrying your application looks as sharp as the product behind it. However you make it, make it - it is the single highest-leverage thing in your application.

What the 12 weeks actually look like

Say you get in. Speedrun is not a lecture series - it is a compressed, high-intensity sprint designed to push a company as far as possible before Demo Day. The shape of it:

• In person in San Francisco, roughly one focused day per week of programming, with the rest of the week spent building. You are expected to be there and full-time on the company.
• Direct access to a16z partners and operators - the point of the day-a-week cadence is high-signal feedback, not busywork. You use the network to unblock growth, hiring, and fundraising.
• $5M+ in credits across cloud, AI model providers, and software - the compute and tooling most AI startups burn cash on, largely covered.
• A relentless push toward traction. The program is oriented around getting to a metric you can stand on at Demo Day, where you pitch to a room of 1,000+ investors in two minutes.

The whole arc rewards teams that convert access and credits into shipped product and growth fast - the same trait the application screens for.

a16z Speedrun vs Y Combinator: which fits you?

If you are AI-native and shipping fast, Speedrun's smaller cohort and bigger check are compelling. If you want the largest network and the broadest brand, YC still wins on scale. Plenty of founders apply to both. For YC specifically, see our YC application tips guide.

5 mistakes that get Speedrun applications rejected

1. Treating the form as the application. No demo link, no video - you are asking a reviewer to imagine your product. They won't.
2. Selling a vision with no shipped product. Speedrun funds runnable ideas. "We will build" loses to "here it is, try it."
3. Hiding your velocity. If you built it in six weeks and have 400 users, that is the headline. Bury it and you look average.
4. Generic "we use AI" positioning. Everyone uses AI now. Name the specific thing AI made newly possible that you are building.
5. A polished deck but a clunky demo. Invert it. A rough deck with an undeniable live demo beats the reverse every time.

The bottom line

Speedrun accepts fewer than 1 in 250 founders, and the terms are worth the odds. But you do not beat that funnel by being comprehensive - you beat it by being undeniable. Ship something real, show it moving fast, and put a two-minute demo at the center of your application instead of a written pitch. In a program built to reward velocity, the founder who shows up with a working demo has already made the argument.

So here is the one-line action: record the demo video today, even a rough cut, and let the product make your case. Watch the Speedrun site for the next application window, apply early because review is rolling, and if you want the intro video to look as sharp as the product behind it, Flowjam turns your screen recording into a launch-ready video in hours - start here. The founders who get in are not the ones with the best decks. They are the ones who shipped, and showed it.

Frequently asked questions

How much does a16z Speedrun invest?
Up to $1M per company: $500K for 10% upfront in a SAFE, plus another $500K in your next round within 18 months, alongside $5M+ in partner credits.

What is the a16z Speedrun acceptance rate?
Under 0.4% - roughly 1 in 250 applicants - with about 60-70 teams accepted per cohort.

How long is the program and where is it?
12 weeks, in person in San Francisco, at roughly one focused day per week plus independent work.

Do I need a co-founder or prior funding?
No. Solo founders are accepted and prior funding is fine. Complementary teams are preferred but not required. You must be able to attend in person; the Global Founders Program helps with visas.

What does Speedrun look for?
In order: velocity and traction (shipped product, real users, growth), founder quality with a genuine domain edge, and an AI-native idea with huge potential.

When is the next cohort?
SR007 runs July 27 to October 11, 2026 (applications closed May 17, 2026). Speedrun runs two cohorts a year and reviews rolling, so watch for the next window and apply early.

Should I apply to both Speedrun and YC?
Yes, plenty of founders do, and the deals do not conflict at the application stage. Speedrun offers a bigger check and a smaller, AI-focused cohort; YC offers scale and the broadest network. If you are AI-native and shipping fast, prioritize Speedrun; if you want the widest brand and network, prioritize YC.

Do I have to move to San Francisco?
You must attend in person for the program, so yes, you need to be in San Francisco for the roughly one-day-a-week programming across the 12 weeks. Your company does not have to be based there, and the Global Founders Program helps international founders with visas.

What if my demo is not polished yet?
A rough but real demo beats a polished pitch for a product that does not exist. Reviewers want to see the thing work, not see it look perfect. Ship what you have and show it running - then tighten the intro video around it.

Do I need to be incorporated to apply?
You can apply early-stage; company details are part of the form, but the bar is a working product and velocity, not paperwork. Sort incorporation as you would for any pre-seed raise - do not let it block your application.

Related reads for founders: YC Application Tips 2026 · Seed Round Valuation Guide (how the SAFE and your next round are priced) · How to Announce a Product Launch