Y Combinator, the accelerator that pioneered the batch model, runs a three-month program four times a year and invests $500,000 in every company it accepts, according to its published program documentation. The standard deal is $125,000 for 7 percent on a post-money SAFE plus $375,000 on an uncapped SAFE with a most-favored-nation clause.
How does the standard accelerator deal work?
Y Combinator's own about page breaks the money into two instruments. As stated in the program documentation on YC's site: "YC invests $500k per company: $125k for 7% on a post-money SAFE and another $375k on an uncapped SAFE with an MFN." The two halves behave differently, and the difference is the whole point.
The first SAFE fixes a price and an outcome on day one: $125,000 buys 7 percent, calculated post-money, so dilution from other investors does not change what the accelerator holds. The second $375,000 converts later at whatever terms the next priced round sets, provided they are no worse, which is what the most-favored-nation clause enforces. Founders get more total capital without surrendering more fixed equity, and the accelerator gets a larger position in the companies that raise well.
The 7 percent figure is the number to underwrite mentally. It is compensation for the program, the network, and the signal of selection, paid in equity rather than fees, and it is permanent in a way program benefits are not. Everything else an accelerator offers is downstream of that trade.
What actually happens during a three-month batch?
The program compresses a company's first year into roughly eleven weeks of concentrated work, and the documentation is unusually concrete about the mechanics:
- Any startup, anywhere in the world, can apply through an open application process, per YC's FAQ.
- Accepted founders relocate, since the batch runs in person in San Francisco, opening with a three-day retreat and weekly meetups.
- Startups are sorted into three groups, each led by YC partners who advise founders in office hours.
- Each group splits into sections of six to ten companies, keeping a small-group setting inside the larger batch.
- The three-month cycle closes, and the alumni network takes over as the durable asset.
The FAQ is explicit that the in-person format is a considered choice: "We briefly did run YC remotely during Covid, but since 2022 it has been back in person. We've found that YC works much better in person," states the FAQ on YC's site. The structure is not decoration; density of contact is the product, and the weekly rhythm exists to force decisions that solo founders defer.
What does the accelerator get in return?
Equity, purchased at a fixed price, in volume. An accelerator's portfolio economics work because most investments fail quietly and a small number return the fund. A fixed 7 percent position across hundreds of companies per year turns individual judgment into a diversified book, which is why the model scales while partner-driven seed funds stay small.
Selection does most of the work before the program starts. The documentation notes that startups arrive at all different stages, some not yet working on anything, which means the batch's value is partly in standardizing the earliest, most erratic phase of a company's life rather than in teaching a secret curriculum.
Why do accelerators exist at all?
The batch solves three early-stage problems at once. It standardizes a first check for teams that lack traction or connections. It compresses advice into office hours with people who have watched hundreds of the same mistakes repeat. And it manufactures a peer group, which the documentation treats as a permanent feature rather than a perk: the alumni community is described on the about page as an increasingly valuable resource that continues long after the three-month cycle ends.
None of this requires believing accelerator graduates outperform everyone else, and no figure on this page claims they do. The honest framing is that a batch converts an unstructured first year into a scheduled one, with the schedule itself as the deliverable.
How does the batch model compare with a seed fund?
The instruments overlap but the services differ. A seed fund prices each deal on its own merits after diligence; an accelerator posts one public price and competes on volume and selection instead. A fund's help is continuous and financial; a batch's help is compressed and social, front-loaded into eleven weeks and then handed to the network.
For a founder, the practical comparison is about fit rather than prestige. Teams that need a structure, a deadline culture, and a first check at speed fit batches. Teams that need a large check, patient capital, and board-level guidance per company fit funds, and the two are not mutually exclusive across a company's life.
There is also a signaling asymmetry worth naming. A batch acceptance is a public, dated endorsement that investors read as pre-screening, while a quiet seed round carries no comparable badge. Founders raising in crowded categories sometimes take accelerator terms precisely for that signal, and it is a legitimate reason, separate from anything the program teaches.
The reverse signal exists too. Declining a batch to stay unlisted suits companies whose edge is confidentiality or whose buyers would not care. The signal only has value where the audience respects its source, and founders are the ones who know their audience.
Who should skip an accelerator altogether?
Founders with clear distribution, committed capital, or a running revenue engine trade the most for the least. The batch model prices its help in permanent equity, and a company that already knows its next three moves pays full price for repetition. The documentation itself frames the program as serving teams at the start of the curve, including some that have not yet begun working in earnest.
A useful stress test is to price the alternatives. The same dilution could buy a senior hire, a year of runway, or a distribution deal, and a batch competes against all three. What the program genuinely monopolizes is time density: eleven weeks of forced contact with partners, peers, and alumni that would otherwise take a year of cold outreach to assemble.
The decision rule fits on one line: take the deal when access and structure are the binding constraints, skip it when they are not, and either way read the standard terms on the accelerator's own page before signing.
