The ARPA-E concept paper requirements fit in a paragraph. Four pages maximum, including every figure, table, and citation. PDF, 12-point Times New Roman, one-inch margins. Four required sections: Concept Summary, Innovation and Impact, Proposed Work, and Team Organization and Capabilities. It is mandatory. No compliant concept paper, no full application.
That is the format. You can satisfy every line of it and still get a discouragement letter three weeks later. Discouraged is ARPA-E's word for no.
The format is not what rejects energy startups. ARPA-E does not fund improvements to your technology. It funds technologies that put your industry on a different cost curve. A 30% better battery is an improvement, and it gets declined no matter how well you wrote it.
So this guide covers the format in two minutes, then spends the rest of its time on the bar that actually decides your outcome: the learning-curve test ARPA-E publishes and almost nobody reads, a four-question test to classify your innovation before you write a word, and the routing call to make when the honest answer is that you belong somewhere else.
What are the ARPA-E concept paper requirements?
The concept paper is a four-page PDF submitted through ARPA-E eXCHANGE. It must address a single technology; bundled concepts get rejected on compliance. The formatting rules are enforced mechanically. From the ARPA-E OPEN funding opportunity announcement:
| Requirement | Specification |
|---|---|
| Page limit | 4 pages maximum, including graphics, figures, tables, and citations |
| Format | Adobe PDF, English |
| Page setup | 8.5 x 11 inches, margins no less than 1 inch, single-spaced |
| Typeface | Times New Roman, black, 12 point or larger (figures and tables exempt) |
| Header, every page | ARPA-E control number, lead organization name, PI last name, upper right |
| Footer, every page | Page number |
| Scope | One concept per paper. No bundling. |
One rule catches people every year: citations count against your four pages. If you planned a page of references, you have written a three-page concept paper.
Your first paragraph is a data block, not prose: lead organization name and location, PI name, technical category, funding requested (federal and cost share), project duration.
Flagging this honestly: page limits and criteria are set per funding opportunity, not agency-wide. The numbers above are from the OPEN FOA. Read the FOA you are actually applying to, and read it before you trust this article.
None of the above is what declines you.
How does ARPA-E evaluate concept papers?
Under the OPEN FOA, the concept paper is scored against two criteria, weighted equally:
- Impact of the Proposed Technology Relative to FOA Targets (50%)
- Overall Scientific and Technical Merit (50%)
Most founders read that, assume Impact means market size, and spend all four pages on Scientific and Technical Merit, where they are comfortable. Half the score lives in a criterion that is not about your science.
Impact is scored on four things:
- Potential for a transformational and disruptive (not incremental) advancement compared to existing or emerging technologies
- Positive impact on at least one ARPA-E mission area
- Identification of the techno-economic challenges that must be overcome for the technology to be commercially relevant
- Awareness of competing commercial and emerging technologies, and how your concept significantly improves on them
Read bullet 3 again. ARPA-E is asking you, at concept-paper stage, to name the economic problems you have not solved. Founders instinctively hide those. The criterion awards points for surfacing them.
Bullet 4 means a paper that does not name its incumbent competitors, with real numbers, leaves half its score partially unaddressed.
A caveat, because this moves: ARPA-E's boilerplate now marks the concept-paper criteria "deleted and reserved for future use" and defers to each FOA. The 50/50 split is what OPEN used. Your FOA governs.
Above the criteria sit ARPA-E's program policy factors, unscored but decisive in close calls. The big one is portfolio balance, which explicitly includes technical and commercialization risk and stage of technology development. A sound paper can be discouraged because ARPA-E already funds three teams doing something adjacent.
Why ARPA-E rejects a 30%-better battery
ARPA-E explains this bar themselves, in a passage almost nobody reads. From the FOA provisions:
Existing energy technologies generally progress on established "learning curves" where refinements to a technology and the economies of scale that accrue as manufacturing and distribution develop drive down the cost/performance metric in a gradual fashion. This continual improvement of a technology is important to its increased commercial deployment and is appropriately the focus of the private sector and it can be spurred by early-stage R&D supported by the applied energy offices in DOE. By contrast, ARPA-E supports high-risk, potentially transformative research that has the potential to create fundamentally new learning curves.
That is the whole test.
Your technology is either a point on the existing learning curve or a new curve. If it is a point on the existing curve, ARPA-E's own charter says you belong to the private sector or DOE's applied offices. Not because the work is bad. Because it is someone else's job.
Now the sentence that reframes everything. ARPA-E projects:
typically start with cost/performance estimates for the proposed technology that are well above the level of the competitive incumbent technology.
Sit with that. ARPA-E expects your technology to be more expensive than the incumbent today. Being worse right now is normal. Being on the same curve is fatal.
Founders have this backwards. They lead with "we are 30% cheaper than lithium-ion," believing it is the strong claim. To a program director it is the disqualifying claim: you are on the lithium-ion curve, and that curve gets there without you.
The numbers back the program director up. BloombergNEF's December 2025 battery price survey put average lithium-ion pack prices at $108/kWh, down 8% in one year, with LFP packs at $81/kWh and stationary-storage packs at $70/kWh. A curve falling 8% a year erases a 30% advantage in about four years. Your project takes three.
So stop asking "is my technology better?" ARPA-E asks: is it on a different curve, and what mechanism puts it there?
The ARPA-E disruptive innovation test: classify yourself A, B, or C
Here is the pre-screen we run on every ARPA-E client before agreeing to draft anything.
To be clear: this is Cada's internal rubric, not ARPA-E's. ARPA-E does not publish an A/B/C scale. We built it because "not transformational" is the most common way strong energy companies lose, and we would rather predict that in ten minutes than discover it in three months.
Every innovation lands in one of three buckets, and the bucket caps how we score disruptive innovation (we score it 1 to 9 internally):
| Class | What it is | Score band |
|---|---|---|
| A | New mechanism or first-principles shift | Floor of 7 |
| B | Novel application of a known mechanism | Floor 5, ceiling 7 |
| C | Engineering optimization of a known approach | Ceiling 4. Reject. Not ARPA-E. |
The caps are the point. A cap is a decision; a score is an opinion. No framing, no drafting skill, no consultant turns a C into an A. If you are a C, the best possible outcome of a beautiful ARPA-E concept paper is a well-written decline.
Four questions decide your class.
1. Does your innovation contain at least one scientifically or engineering-novel element? A new chemistry, catalyst, or electrolyte; a new device architecture; a new thermodynamic operating regime; a new materials process or fabrication route; a new control or power-electronics topology; or a techno-economic operating point reachable only through your innovation.
If yes, name it and cite the prior work it goes beyond. That element becomes your concept paper. If the honest answer is "we integrate existing components in a new way," you have to earn it through the next three.
2. What two or three specific failure modes does your integration resolve that prior integrations did not? Named technical or techno-economic failures. Not "gaps in the market."
3. Why has no prior group done this integration? Acceptable answers: a component that did not exist until recently, a manufacturing route that did not exist, an economic disincentive at the incumbent's current $/kWh, a materials supply constraint. "Nobody thought of it" is not an answer, and a program director will not believe it.
4. What technical risk does doing this for the first time introduce? This goes into Proposed Work as a named risk with a mitigation.
If you cannot answer question 1, and cannot answer 2, 3, and 4 with substance, you are a C. Stop. Do not write the paper.
Three fictional examples:
Class C. A team redesigns pack-level thermal management and gets 30% more usable energy density from standard NMC cells. Real engineering, real value. Same chemistry, same curve, and the cell manufacturers are working on it too. Ceiling 4.
Class B. The same team takes a solid-state electrolyte demonstrated only in coin cells, moves it into a stationary-storage form factor nobody has, and pairs it with a fabrication route that makes it manufacturable at scale. Known mechanism, novel application. Floor 5, ceiling 7, and genuinely fundable.
Class A. The team proposes a new electrochemical couple whose theoretical cell voltage puts a floor under $/kWh that lithium chemistry cannot physically reach, backed by a coin-cell result and the thermodynamic argument for that floor. New curve. Floor 7.
A is rare. Most funded ARPA-E work is B, and B is fine. The line that matters is B versus C.
ARPA-E vs DOE EERE: which one is actually yours?
Failing the A/B/C test is not a death sentence. It is a routing error, and routing errors are cheap to fix before you write.
ARPA-E funds research that creates a new learning curve. DOE's applied energy offices, including EERE, fund moving along the existing curve and scaling it: applied R&D, pilots, demonstration, deployment. The same technology is often strong at one and disqualified at the other.
ARPA-E's boilerplate says so outright. Continual improvement of an existing technology "can be spurred by early-stage R&D supported by the applied energy offices in DOE." They are telling you where to go.
| Dimension | ARPA-E | DOE EERE and applied offices |
|---|---|---|
| Funds | A fundamentally new learning curve | Progress along, and scaling of, the existing curve |
| Your cost today | Expected to be above the incumbent | Expected to be competitive, or heading there |
| Rewards | A mechanism the incumbent cannot reach | Efficiency, cost reduction, manufacturability |
| Disqualifies you | Incremental improvement | High-risk science with no deployment path |
That Class C thermal-management team is genuinely strong for an applied-office program or a DOE SBIR, and a guaranteed decline at ARPA-E. Same technology, different door, opposite outcome.
Honest uncertainty: program availability moves. OPEN has run on a roughly three-year cycle (2009, 2012, 2015, 2018, 2021, and Vision OPEN 2024), and focused programs rotate constantly. Check ARPA-E eXCHANGE for what is actually open.
ARPA-E technology to market: the bar founders underestimate
One correction first, because being confidently wrong here is worse than being vague.
In ARPA-E's standard full-application criteria, technology to market is not a separate weighted criterion. The four criteria are Impact of the Proposed Technology (30%), Overall Scientific and Technical Merit (30%), Qualifications, Experience, and Capabilities of the Project Team (30%), and Soundness of Management Plan (10%).
Commercialization lives inside Impact, as its fourth bullet: "a reasonable and effective strategy for transitioning the proposed technology from the laboratory to commercial deployment."
That is harder than having its own criterion, not easier. Your commercialization story is scored in the same 30% block as your disruptive-innovation claim. A weak technology-to-market answer does not cost you a self-contained slice of points. It drags down the criterion carrying your reason for existing.
It is a hard gate at the end, too. Before an ARPA-E award executes, the recipient must submit a Technology-to-Market Plan and obtain the program director's approval. Commercialization is never a later problem here.
Four things move the technology-to-market score:
A named T2M lead. A real person who owns commercialization. Not "TBD," not the CTO wearing a second hat. A proposal without one reads as academic, and academic is a category, not a compliment.
Named customer segments. Not "the grid." Not "industry." Name them: distribution cooperatives, behind-the-meter commercial and industrial operators, IPP offtakers, utility-scale storage developers. Specific enough that a program director could call one.
Unit economics at scale, anchored to a tier-1 source. Lazard, EIA, IEA, NREL, BloombergNEF. Your $/kWh at manufacturing scale against a published incumbent baseline, not against competitor startups whose numbers are marketing.
The techno-economic challenge you have not solved. Name it. The Impact criterion literally asks for it. Founders bury it, and program directors read the omission as naivety.
A 10-minute self-screen before you write a word
- Name your incumbent and cite its number from a tier-1 source. If you cannot, stop.
- State your hero metric against that number: $/kWh, $/GGE, $/tonne CO2, percent efficiency, LCOE.
- Answer the curve question. A point on the incumbent's curve, or a different curve? Name the mechanism.
- Classify yourself A, B, or C. If C, go to EERE or a DOE SBIR today.
- Name your T2M lead. If the answer is "TBD," you are not ready to write.
- Name three customer segments specifically enough that a program director could call one.
- Name the techno-economic challenge you have not solved.
- Check the ARPA-E project library for overlap. Program directors know their portfolio cold. If you do not raise an overlapping funded project, they will.
Ten minutes. The alternative is 60-plus hours on a paper that was never in scope, and a discouragement letter that never says why.
Before you spend 60 hours on the wrong paper
The A/B/C test is the same pre-screen we run internally, and you can run it on yourself this afternoon for free. If it says you are a C, you just saved a quarter.
What Cada sells is the execution after the diagnosis: the one-page concept doc that gets a program director to take a call, the outreach strategy behind it, and the concept paper itself. Cada has written 100+ proposals across 30+ agencies, including the energy portfolio most grant consultants do not touch.
If you want a second opinion before you commit, book a free ARPA-E fit check. Bring your technology and your hero metric. Twenty minutes, and you leave with a straight A, B, or C answer and a routing recommendation, whether that is ARPA-E, EERE, or a DOE SBIR. Sometimes the honest answer is do not apply, and we will say it. No pitch, no obligation.
The ARPA-E concept paper requirements will take you an afternoon to satisfy. Deciding whether to write one at all is the ten minutes that actually matters.
Page limits, evaluation criteria, and program availability are set per funding opportunity announcement and change between cycles. Always verify against the FOA you are applying to before you write. Sources: ARPA-E OPEN Funding Opportunity Announcement (DE-FOA-0002459), Sections IV.C and V.A; ARPA-E FOA Provisions Incorporated by Reference (December 2024), Sections I.A and V.A; BloombergNEF Lithium-Ion Battery Price Survey, December 2025; Cada's internal ARPA-E 1-Pager Playbook.
