Half-Backed
For projects too good to stay side projects.
- CLOUD
- $100K
- compute and resources, per project
- prize
- $1M
- for the best project
- when
- 20-22mar
- Zürich · J floor · 2027
We back projects that already have code, prototypes, and results.
But no lab, company, or grant to push them forward.
Selected projects get compute and resources, and a week in Zürich to bring them to the next step.
Apply now with one markdown file.
The gap
Grants fund research. Accelerators fund companies. Employers fund roadmaps. Half-Backed exists for projects that fits none.
We back promising projects stalled when you changed labs, the team moved on, or the next step needed more compute or expertise than you could get.
What qualifies
Bring
- Evidence of prior work: code, a prototype, an experiment, a simulation, or a preprint.
- A specific technical question or next step that would materially change what you believe about the project.
- An honest account of why it's stalled: compute, infrastructure, time, missing expertise, a collaborator, or a lack of institutional fit.
- Work that is not already properly backed by your employer, primary research agenda, or startup.
Do not bring
- A bare idea, a pitch deck, or a request for startup capital.
- Something a normal hackathon could finish.
- A company experiment that can be funded internally.
- An API wrapper with no technical question underneath it.
What we back
- Up to $100,000 per project in compute to unblock your work.
- Dedicated resources. Before the sprint, we scope the resources, setup, and missing pieces each selected project needs.
- A week in Zurich. 20-22 March 2027, Zürich. Unblock your project and present the results in front of potential new collaborators, hires, and investors.
We take no equity and no IP. Zero fees.
The prize
$1,000,000
a SAFE backed by top VCs.
Participating teams unblock their projects. One leaves fully backed.
yours to decline
How this runs
Send one Markdown file (see below). Tell us what you have already done, what you are trying to find out, and what has stopped you.
We choose a small global cohort for relevance, impact, technical depth, and evidence.
Before Zürich, we scope the resources, infrastructure, and people each project needs.
At J floor, run the planned work and present the results. You may find more support, team members, or investors — and one of you leaves with the $1M SAFE.
Apply
Just one Markdown file.
As clear and short as possible.
Rolling review. We close when the cohort is full.
--- project: "" # one line, no adjectives team: [] # names & emails; solo is fine contact: "" # email we reply to based_in: "" # city, country stuck_since: "" # first commit, notebook, or prototype blocker: "" # compute | expertise | collaborator | data | infrastructure | time ask_usd: 0 # estimated resource cost; ceiling is $100,000 resources: "" # what you need, or "see resource plan" receipts: [] # repo, preprint, demo, notebook, benchmark --- ## What are you trying to find out? Three paragraphs, for a sharp reader outside your field. ## What have you already done? Link the evidence: commits, runs, measurements, a prototype. ## Why is it stuck? The honest reason — and why your employer, lab, or grant can't cover it. ## The decisive next step The experiment, build, or analysis that would materially change your view. ## What would change your mind? The outcome that would make you stop, change direction, or commit to the next stage. ## Resource plan What you need, what it costs, and what makes it technically feasible. ## Who or what is missing? Specialist expertise, a collaborator, infrastructure, or anything else that would move it.
curl -O halfbacked.org/half-backed.md
Questions
What does "up to $100,000" mean?
Each selected project gets access to up to $100,000 in compute resources, given by our partners and sponsors. The amount you receive will depend on what you need.
What's the $1M SAFE?
The best project gets a $1M investment as a SAFE from top VCs. If you'd rather not take it, you can say no.
What if the project does not work?
Too bad.
Do I need a startup?
No. We do not require business plans. The work might lead to a paper, open-source project, company, or a settled question.
Can I apply solo?
Yes. Individuals and teams can apply, but, if you need help, make sure to invite also collaborators.
Do you take equity or IP?
The program takes no equity and no IP.
Who reviews applications?
Applications are reviewed confidentially by the organising team. Further reviewer details will be announced.
halfbacked.org/apply/ · upload or curl