Unquestionably, one of many hottest new startup accelerators in tech proper now could be Andreessen Horowitz’s Speedrun program. Launched in 2023, the accelerator has an acceptance fee of lower than 1%. In a January blog post, this system mentioned that over 19,000 startups pitched and fewer than 0.4% have been accepted into the newest cohort.
This system used to deal with gaming startups, then expanded into leisure and media, and is now a “horizontal program,” Joshua Lu, this system’s normal supervisor and a companion at a16z, advised TechCrunch. At present, founders of any kind of startup can apply, and this system runs for about 12 weeks in San Francisco. It as soon as had a program in Los Angeles, however Lu mentioned the main focus shall be on SF any further.
There are two cohorts a 12 months, and round 50 to 70 startups are accepted into every. This system invests up to $1 million into every firm, although the draw back is that it’s a bit dear. It usually invests $500,000 up entrance in change for 10% of the startup’s firm by way of a SAFE be aware, and one other $500,000 if the following spherical is raised inside 18 months, at no matter phrases agreed to by the opposite traders.
As compared, Y Combinator typically takes a fixed 7% of the corporate for $125,000, with one other $375,000 “invested on an uncapped MFN secure.”
Speedrun mentioned its program is extra “fairness costly” due to what it provides founders. It supplies them with entry to a16z’s advisory and enterprise networks that help with duties like go-to-market, model growth, media technique, and expertise sourcing. Plus it provides the startups perks like $5 million in credit to distributors similar to AWS, OpenAI, Nvidia, and Deel.
Given the excessive curiosity, and low acceptance fee, TechCrunch spoke to Lu for some recommendations on how startups can finest stand out. The most recent cohort started in January and can finish in April with a Demo Day. Functions for the next cohort open in April, although it seems to be at low season purposes year-round, Lu mentioned.
Concentrate on the founding crew
Speedrun focuses on early-stage startups. Due to this, they actually look at who’s on the founding crew and whether or not their expertise complement one another, Lu mentioned.
Techcrunch occasion
Boston, MA
|
June 23, 2026
“That doesn’t imply one needs to be technical and one needs to be industrial and one needs to be advertising,” Lu mentioned. It signifies that “we choose to not see any obvious holes in capabilities or pursuits. We wish the founding crew to be self-aware and for that to be a part of the hiring plan.”
In addition they prefer to see groups which have labored collectively earlier than or have a shared historical past.
“There are many issues {that a} founding crew has to navigate of their startup journey and having a little bit of sample recognition, having the ability to work with one another, understanding how you can disagree and how you can come out the opposite facet of a disagreement, these are all issues individuals on founding groups with shared histories have a better time with, on common,” he continued.
Despite the fact that AI has lowered the limitations to constructing software program, it’s nonetheless extremely useful for a founding crew to be technical, Lu mentioned. On the identical time, as a result of AI has made it a lot quicker to construct and validate hypotheses and get a product on the market, Lu mentioned the Speedrun crew likes to see when a startup already has somewhat little bit of market validation or traction for his or her product.
“Speedrun as a program is de facto nice at serving to groups pour gasoline on a really small spark or hearth,” he mentioned. “We search for groups which have endeavored to construct and attempt to present us that there’s somewhat spark we will fan the flames on.”
Restrict the market “idea”
Lu mentioned one frequent mistake founders usually make within the software course of is spending an excessive amount of vitality speaking concerning the market idea or why there’s a outlined downside and why their answer is the best one. “All of which may be true,” he mentioned.
On the identical time, he added, even the largest, most profitable tech firms confronted surprising blockades after they have been younger, typically even pivoting fully. What an organization thinks it’s going to construct initially isn’t essentially what’s going to make it profitable on the finish.
“What we actually wish to hear about is why this founding crew is de facto good collectively,” he continued, “why they’re a terrific founding crew, the absolute best founding crew to resolve this explicit downside.” After which on high of that, any validation on the thought itself.
It’s okay to make use of AI for the applying, however…
Lu mentioned this system encourages each founder to make use of AI to “clear up” their software. He mentioned there’s now no excuse for grammar errors or misspellings given the rising sophistication of AI instruments. He additionally mentioned AI may help founders kind out their ideas, making them clearer, extra concise, and extra coherent.
But when AI did all of the work in explaining the startup, that will backfire. If a founder makes it to the following spherical, will probably be a reside video-call interview. “At that time, their reside narration clarification expertise are going to be put to the take a look at,” he mentioned. So founders needs to be ready to speak cogently about their startup with out the assistance of AI.
Solely about 10% of founders make it to the video-call stage. There are usually two to a few traders on the judging panel at a time.
After the reside interview, the crew usually conducts just a few extra screening calls with the founders, after which a ultimate choice on the cohort is made.
Be grasping to community
There are, in fact, different accelerator applications for startups to select from. Lu mentioned Speedrun itself was impressed by a few of these different applications.
Nonetheless, he mentioned, this accelerator prides itself on giving founders entry to a big, specialised working crew. The truth is, he mentioned the perfect groups that get essentially the most out of this system are those most “grasping about getting publicity to the superb individuals and applications” Speedrun has to supply.
Lu listed off only a few factors: a16z has round 600 individuals, and 10% of that workers is on the funding crew, he mentioned; everybody else is an operator who helps the businesses the agency works with. In consequence, founders in Speedrun could have entry to specialists who may help with advertising, banking, finance, administration, and lots of different capabilities. So it helps to know who the startup needs to attach with and why.
“We inform founders that come via this system, what you get out of Speedrun is what you place into it,” he mentioned. “We expect founders who wish to make the most of world specialists in many various domains early of their startup journey could be actually good to decide on us.”
Recommendation from a founder in this system
Founder Mohamed Mohamed, who’s within the latest cohort, just announced a $5 million raise for his proptech startup Sensible Bricks led by a16z’s Speedrun. He was drawn to this system as a result of he mentioned it stood out as one of many few “explicitly designed for co-founders engaged on frontier AI purposes,” and he picked it as a result of he needed a program that will permit him to “stress-test an formidable technical imaginative and prescient.”
Mohamed mentioned he handled the applying like an inside technique memo moderately than a pitch. “As an alternative of sharpening buzzwords, we centered on readability — the actual downside, why it’s structurally onerous, and why our crew is unusually well-positioned to resolve it,” he mentioned. “We have been specific about what was working, what wasn’t, and the place we would have liked assist. I feel that honesty and clear articulation of why this downside issues” is what helped the corporate within the software course of.
He referred to as the entire course of “rigorous however refreshingly considerate,” and mentioned it was designed to grasp how founders assume, not simply what they’ve constructed thus far. “The conversations went deep into product structure, information technique, and long-term ambition. It felt nearer to a partner-level dialogue than a typical accelerator interview, which was a powerful sign for us,” he mentioned.
His general recommendation is to be “intellectually trustworthy and exact.” For instance, he mentioned in his software he prevented “over-optimizing” for the sake of hyping up his firm. “When you’re imprecise, spinoff, or overly defensive about your thought, it exhibits shortly. Don’t attempt to sound larger than you might be; readability about the place you truly are is way extra compelling than inflated narratives,” he mentioned.
In the long run, “Speedrun isn’t in search of good firms; they’re in search of founders who can cause clearly about advanced issues and construct with conviction,” he mentioned. “Articulate the onerous elements of what you’re doing and why they’re value tackling. Depth beats polish each time.”
Correction, story initially misstated YC’s funding for its 7%. It has been corrected.
