Grant Management Associates — Funding Intelligence. Winning Strategy.

On May 7, the Department of Energy's Office of Critical Minerals and Energy Innovation announced $52 million for 20 projects across 15 states. The headline framing was competitiveness, onshoring, and trade exposure. The projects themselves were laser curing systems, poultry rendering, molten sulfide ironmaking, ceramic membranes, and cement made from lithium battery waste.

Most people skim the summary and move on. That's a mistake, because a selection list is the most honest document a federal program office ever publishes. A funding opportunity tells you what an agency says it wants. A selection list tells you what it actually funded.

I worked through the underlying project records — leads, partners, cities, dollar figures, technical scope. The patterns are consistent enough to function as instructions. Here is what the winners share, what activities inside those awards are fundable, and what any company, utility, or tribal government should take from it.

Start with the sizing, because it sets your strategy

Awards run from $1,203,971 to Lawrence Technological University for low-energy laser curing of powder coatings and porcelain enamel, up to $5.6 million to MIT for iron production by molten sulfide electrolysis with Rio Tinto. Median and mean both land at roughly $2.65 million.

The more useful detail: $3,000,000 appears exactly, six separate times — the Institute of Gas Technology, SUNY Buffalo, Sortera Technologies, Washington State University, Ammobia, and E2H2NANO. Six identical figures out of twenty is not coincidence. That is a ceiling applicants were writing to.

It is not absolute — the University of Minnesota came in at $3.1 million and MIT at $5.6 million — but it tells you where the mass of the distribution sits. Scope your project to three million dollars over roughly three years unless you have a specific reason to exceed it. Applications that blow past the cap force reviewers to either cut your scope or pass, and cutting scope is more work than passing.

What the winners have in common

1. Every project attaches to an import or a domestic supply constraint.

Ammobia, Inc., of San Francisco won $3 million for a single-pass ammonia reactor operating at roughly ten times lower pressure and 150°C below conventional processes, targeting a twofold cut in capital expenditure for one of the world's highest-volume chemicals. Oregon State University won $1.8 million to turn delithiated aluminosilicate — a byproduct of lithium battery production — into supplementary cementitious material, with Albemarle and Amrize as partners. The University of Texas at Austin won $2.37 million for membranes that separate high-value unsaturated hydrocarbons, the feedstock for polymers, petrochemicals, fuels, and food production.

The framing is consistent: this thing we currently import, or can't source domestically at volume, gets made here instead. If your technology touches an imported material, quantify it. If it doesn't, find the input in your process that is imported and build the case there.

2. Improvement is stated as a delta against the incumbent, in numbers.

Wilson Engineering Technologies of Pleasant Hill, California won $2.3 million for an Energy Flows Redistribution Process for poultry rendering, with Tyson Foods anchoring. The claims: 45% less energy waste per unit of output, 125% more production capability, 45% lower operating expense. Not one is an absolute performance figure. Every one is measured against the process being displaced.

Same discipline throughout. WPI: up to 40% energy savings in food, 20% in paper. Lawrence Tech: curing times cut from minutes to seconds. One iron and steel project targets 25% lower energy demand in direct-reduction pellet production with 3–5% higher iron recovery.

The cleanest example is the University of Tennessee's wastewater aeration project: 70% less energy for aeration, 30% less for mixing, and at least 50% overall energy and cost savings measured explicitly against conventional activated sludge — the process actually running in most treatment plants today. Named incumbent, named baseline, numbers attached.

Reviewers score what is in front of them. They do not build your comparison for you.

3. Waste streams become domestic feedstocks.

Lithium battery byproduct becomes cement. Poultry rendering byproduct becomes usable material. The University of Maryland's $1.47 million 3D-printed polymer composite heat exchanger captures low-grade waste heat from sources including data centers. Washington State University's peroxyacid pulping work recovers low molecular weight organic acids — widely used chemicals — from a process that currently just burns energy.

This one pattern hits three priorities simultaneously: it cuts energy demand, substitutes for an import, and creates domestic supply. If you generate a waste stream, you may be sitting on a fundable project you haven't recognized.

Bluestem Biosciences of Omaha adds a variation worth copying. Its $2.4 million project ferments underutilized domestic biomass — corn and agricultural byproducts — into ethyl acrylate, targeting the $9 billion paint, coating, plastic, and personal care market at 40% lower energy cost. Two details make it strong: the feedstock creates new demand for American farm output, and the process is designed to run in existing infrastructure. Retrofitting assets that are already built is cheaper, faster, and easier to finance than greenfield construction, and reviewers know it. If your technology can be dropped into installed capacity, lead with that.

4. AI appears as instrumentation on a physical process, never as the product.

Sortera Technologies of Markle, Indiana won $3 million for AI-driven modular sorting that removes copper contamination from steel scrap, with Purdue, MIT, Penn State, and ThermOhm. WPI is building physics-based AI and machine learning frameworks coupled to fiber optic sensors, running a pilot-scale dryer testbed. Missouri S&T's blending platform sits on a materials database.

In every case the model is bolted to a specific industrial problem with measurable physical output — and carries a hard performance threshold. Missouri S&T isn't promising to "apply AI to cement." It commits to 90% predictive accuracy across blends at 60–80% substitution, built on more than 20,000 data points. Applications that lead with the algorithm rather than the plant, or that describe capability without committing to a number, tend not to survive technical review here.

5. A named beachhead plus a credible expansion path.

The Institute of Gas Technology won $3 million for a radiative heat exchanger preheating combustion air to 500°C in corrosive exhaust up to 900°C — named applications being glass melting and aluminum die casting. Lawrence Tech names dish racks and specialty coatings alongside the general powder-coat market. Ames Laboratory's elastocaloric work names ethanol distillation specifically.

One concrete entry application, then the adjacency. A platform with no beachhead reads unfocused; a point solution with no expansion path reads too small to fund.

DOE funded the same problem twice, on purpose

This is the finding I'd put in front of any executive who has ever said "someone's already doing that, so there's no point applying."

Look at copper contamination in steel scrap — a specific, well-known defect that limits how much recycled scrap American mills can put into high-value steel. DOE funded two completely different attacks on it. MIT took $5.6 million to convert ore and scrap to iron sulfide and reduce it electrolytically to molten iron, a route that strips copper in a way the description says no existing process can. Sortera Technologies of Markle, Indiana took $3 million for a two-stage system: AI image-recognition sorting of shredded scrap, followed by a thermo-mechanical process that selectively reacts the copper to enable mechanical separation.

The same thing happened in ammonia. Ammobia of San Francisco received $3 million for a novel reactor running at roughly ten times lower pressure and 150°C below conventional conditions, partnered with Lawrence Berkeley National Laboratory. E2H2NANO of Buffalo received $3 million for a compact modular membrane reactor that integrates reaction and separation to get past the thermodynamic conversion limit, partnered with Johnson Matthey, SUNY Buffalo, and the University of South Carolina.

And again in cement. Oregon State University took $1.8 million to qualify one specific new material — delithiated aluminosilicate, a byproduct of lithium battery production — as a supplementary cementitious material, with Albemarle and Amrize. Missouri University of Science and Technology took $2 million to build the AI platform that qualifies such materials generally, predicting performance for concrete blends at 60–80% substitution against a database of more than 20,000 data points. One project goes deep on a single feedstock; the other builds the tool for evaluating many. Both address the same barrier: getting clinker out of American cement using domestic waste.

Three problem areas, six projects, roughly $18.4 million. Program offices are not trying to pick the winner. They are buying multiple shots at problems they have decided matter, because they don't know which approach survives scale-up either.

That inverts the instinct most companies have. Evidence that DOE has already funded work in your problem space is a warmer signal, not a closed door. It means the office has internally justified the priority, has reviewers who understand the technical landscape, and has a portfolio it wants to diversify. What gets you excluded is not sharing a problem with someone else — it's sharing an approach without a differentiator.

Notice too that MIT is a named partner on Sortera's project while leading a competing route to the same defect. Appearing on someone else's team does not disqualify you from leading your own. The ecosystem is smaller and more collaborative than outsiders assume.

E2H2NANO is also worth a second look for a different reason: it is a small LLC holding a $3 million award with Johnson Matthey, a global catalyst company, in a supporting role. The prime does not have to be the largest organization on the team, and small companies should stop assuming they must subcontract to a multinational to be credible.

Win twice with the same technology

Georgia Tech Research Corporation appears twice in this tranche, for a combined $4.83 million, using the same underlying material.

The first award, $2.13 million, develops continuous roll-to-roll manufacturing of reduced graphene oxide nanofiltration membranes in water-based solvents — the making of the thing, explicitly framed around onshoring membrane production to American facilities. The second, $2.7 million with Rayonier Advanced Materials and Mott Corporation, deploys high-flux reduced graphene oxide membranes to dewater kraft black liquor in a single-pass slipstream at a working pulp mill, replacing thermal evaporation in what the project describes as the most energy-intensive operation in the mill — the using of the thing, in a named industry at a named site.

Neither application competes with the other. Each strengthens the other's case. Together they answer the two questions a reviewer always has about a platform technology: can you make this at volume, and does it work in a real plant?

If you have a platform, this is the template. Split it. Write one project on the manufacturing route and one on a specific high-value application with an industrial host attached. Submit both. You are not diluting your odds — you are answering two objections separately instead of cramming both into a single three-million-dollar scope.

SUNY Buffalo did a version of the same thing, leading a $3 million ceramic membrane award while partnering on E2H2NANO's membrane reactor. IPG Photonics turns up on two separate laser projects. The pattern is consistent: genuine depth in a narrow technical area produces multiple bites in a single competition. Breadth does not.

The case for being a partner

Rayonier Advanced Materials never leads a project in this tranche. It appears as a partner twice — on Georgia Tech's black liquor membranes, and on Washington State University's $3 million peroxyacid pulping project, which targets a 50% cut in energy intensity by replacing the prehydrolysis kraft process for dissolving pulp.

Consider what that company just acquired. Two federally funded attacks on its own energy costs, at two different unit operations in its own mills. No prime administrative burden, no recipient reporting obligation, no cost share as lead. First look at both technologies and a seat at the table while they develop.

If you operate an industrial facility, a utility, or a treatment plant, that is what's actually on offer when a technology developer asks you to join their team. Most operators treat the request as a favor they're doing someone else. It isn't.

You do not need to have invented anything

The University of Minnesota won $3.1 million, with US Steel and a national laboratory as partners, to develop a beneficiation flowsheet that upgrades lower-grade taconite ore to direct-reduction grade — cutting energy demand for DR-grade pellet production by 25% and raising iron recovery three to five percent.

Read the technology list: High Pressure Grinding Rolls, Vertical Stirred Mills, Hydrofloat, Jameson Flotation. Every one of those is commercially available equipment you could order from a vendor this afternoon. Nothing in the project is a new machine.

The innovation is the flowsheet — how the equipment is sequenced, configured, and operated together. And the deliverable is an engineering design, not a piece of hardware.

I raise this because it dissolves the most common objection I hear from operators, utilities, and mid-size manufacturers: we don't have a patent, we don't have a lab, we haven't invented anything, so federal R&D money isn't for us. Minnesota's project is a $3.1 million counterexample. Novel integration of existing commercial technology, aimed at domestic ore bodies that are currently uneconomic, with the energy and recovery gains quantified.

If you operate a complex process, you almost certainly know where it wastes energy and why the obvious fix hasn't been tried. That knowledge is the asset. Pair it with an equipment vendor and a university or laboratory that can model and validate the configuration, and you have the shape of a fundable project — without inventing a thing.

The theme nobody named

Here is something the announcement doesn't tell you and the topic areas don't reveal: more than a third of this portfolio is about separations.

Five membrane projects were funded. The University of Texas at Austin took $2.37 million for facilitated transport membranes that pull high-value unsaturated hydrocarbons out of feedstock streams. SUNY Buffalo took $3 million for carbon-doped titanium oxide ceramic membranes that survive harsh solvent and temperature conditions. Georgia Tech took $2.13 million for roll-to-roll manufactured graphene oxide membranes, and another $2.7 million to apply them to black liquor dewatering at a pulp mill. E2H2NANO took $3 million for a membrane reactor that performs separation and reaction in the same device.

Add Ames Laboratory attacking ethanol distillation and Ammobia targeting ammonia separation and purification, and seven of twenty projects are aimed at the same underlying target. Thermal separation — distillation, evaporation, drying — is one of the largest energy loads in the process industries, and DOE is buying alternatives to it aggressively without having declared a separations topic area anywhere in the solicitation structure.

This is what a selection list gives you that a solicitation cannot. If you work in separations, filtration, or membrane technology, you have a warmer market at this office than the published topic structure suggests.

Two of these carry a second lesson. Georgia Tech's project explicitly onshores manufacture of the membrane itself — not just the industrial process the membrane serves. That's double onshoring: the enabling component and the end product both come home. And both Georgia Tech and Maryland name a specific high-volume production route in the project scope — continuous roll-to-roll for one, 3D printing for the other.

How you would actually manufacture the innovation at scale is being evaluated, not just whether it works in a lab. If your application has no answer to "and then how do you make a million of these," you have left points on the table.

The teaming question, answered precisely

Team size clusters at four to six named partners. WPI carries six: the University of Illinois, the RAPID Manufacturing Institute, Reading Bakery Systems, EPRI, the Alliance for Pulp and Paper Technology Innovation, and IPG Photonics. Ames Laboratory carries six, including the ethanol producer Southwest Iowa Renewable Energy as its host. Lawrence Tech carries three, but they are IPG Photonics, PPG Industries, and Whirlpool.

But four winners had no named partners at all: the University of Texas at Austin, SUNY Buffalo, Georgia Tech, and the University of Maryland. Three membrane projects and one heat exchanger material.

That is the actual rule, and it's more useful than "always team up." When your deliverable is a material or a component at lower technology readiness, a strong single institution can carry it — every solo winner in this tranche was developing a material. When you claim a demonstration under real operating conditions, you need the facility, and the facility has to be on your team. Wilson needs Tyson. Ames needs an ethanol plant. WPI needs a bakery equipment manufacturer.

Ask yourself honestly which one you're proposing. Companies routinely write demonstration-grade promises with materials-grade teams, and reviewers catch it every time.

One more thing worth noticing: IPG Photonics appears on two separate awards, WPI's and Lawrence Tech's. Positioning yourself as the enabling component supplier that shows up across many teams is a legitimate and badly underused federal strategy. You carry none of the prime's administrative burden and build past performance across multiple programs at once.

Note also that Ames Laboratory is a lead, not just a partner. National laboratories both lead projects and support them — Lawrence Berkeley, Oak Ridge, and Idaho National Laboratory all appear on partner lists.

The finding that should change your outreach list

Look harder at who shows up as partners.

The University of Tennessee at Knoxville won $2.5 million for nanobubble-enabled aeration at water resource recovery facilities. Its partners: Carollo Engineers, Oak Ridge National Laboratory — and the First Utility District of Knox County. A public water utility, named on a DOE industrial R&D award.

Missouri S&T's cement supply chain project lists Cleveland-Cliffs, Reserve Management Group, Ecocem Americas, Wiss Janney Elstner Associates — and the Quapaw Nation. A tribal government, named on a DOE industrial R&D award.

This matters and almost nobody is acting on it. Municipal utilities, special districts, and tribal governments generally assume DOE industrial funding is for manufacturers and universities. It isn't. These programs need operating facilities where technology gets validated under real conditions, and a wastewater plant or a tribal industrial enterprise is that facility. The utility contributes site access, operational data, and staff time — much of which can count toward non-federal cost share — and gets first look at a technology that may cut its energy bill.

If you run a water or wastewater utility, you are a valuable partner to a technology developer and probably don't know it. If you are a technology developer without a host site, that is your critical path, and the call you need to make may not be to a manufacturer.

And the ask is smaller than you think

Look at what the Tennessee team actually committed to: a pilot at 200 liters per day, run for six weeks, at a water resource recovery facility.

Two hundred liters a day. For a $2.5 million award.

This is the most encouraging detail in the entire tranche and the one most likely to be misread. Companies routinely talk themselves out of applying because they assume a federal demonstration means standing up something enormous they can't resource or insure. It doesn't. What made this fundable was not scale — it was precision. A named throughput, a named duration, a named facility type, a named incumbent process to measure against, and a stated threshold the project has to clear to count as a success.

A vague promise to "demonstrate at commercial scale" is weaker than a modest, exactly specified validation campaign. Reviewers are trying to assess whether you will actually finish. Specificity is how you prove it.

The fundable activities you should be budgeting

Most companies assume federal funding covers hardware and that everything else is overhead they absorb. The opposite is closer to true. These are allowable, budgetable line items, and several are effectively mandatory:

Techno-economic analysis and life-cycle assessment. DOE's industrial solicitations have consistently required that energy, emissions, and cost claims be validated during the period of performance. This is scored work.
Preliminary front-end engineering design. The FY24 industrial solicitation named Pre-FEED studies explicitly. Pre-FEED is expensive, it's the bridge between a concept and a financeable project, and DOE will pay for it.
Baseline energy characterization at a host facility. This is how you earn the right to claim a percentage improvement.
Testbed construction and test campaigns — at lab scale, as with Ames, or pilot scale, as with WPI. Both were funded in the same tranche.
Sensor deployment, instrumentation, and real-time process control development.
Materials characterization and standards qualification. For anything entering cement, concrete, or structural use, ASTM qualification is a real workstream. Budget it.
Process flowsheet and integration engineering. The University of Minnesota's entire deliverable is a design, not a device. Configuration studies of existing commercial equipment are fundable work.
Manufacturing process development for the innovation itself — Georgia Tech's roll-to-roll membrane production line is a funded workstream, not an afterthought.
Feedstock availability and supply chain mapping.
Database and model development supporting a physical outcome.
Third-party validation at a national laboratory.
Workforce and technician training tied to operating the technology.

If your budget holds only engineering labor and equipment, you have built a weaker application than the ones that won.

The vocabulary moved. The work didn't.

These selections span cross-sector technologies — electrification of industrial heat, efficient energy use in industrial systems, organic wastewater and wet waste treatment — and energy-intensive industries: chemicals and fuels; iron and steel; food and beverage; building and infrastructure materials including cement, concrete, asphalt, and glass; and forest products.

Compare that to DOE's industrial solicitations from three years ago. Same industrial heat. Same chemicals. Same iron and steel. Same food and beverage. Same cement and concrete. Same forest products. The office has been renamed twice and the strategic language has shifted from decarbonization roadmaps to energy dominance and import substitution. The technical priorities underneath are close to unchanged, and the portfolio has actually broadened — asphalt, glass, and wet waste treatment are newer additions.

There is a near-perfect illustration sitting inside this tranche. Ames Laboratory's elastocaloric project — $2.6 million, six partners — was described in an earlier DOE announcement as a solid-state elastocaloric heat pump advancing industrial decarbonization. In this announcement, the same lead, the same funding, and the same partner list appear as elastocaloric process heating for ethanol distillation, framed around energy consumption and industrial competitiveness.

Same engineering. Same team. Same money. Different narrative, eighteen months apart.

That is the whole lesson. Companies that re-engineer their positioning every time the vocabulary shifts burn enormous effort and end up with a folder of half-finished narratives. Reframing is a writing task. Rebuilding is an engineering task. Know which one you actually need — and notice that DOE itself does the reframing when the underlying work is sound.

Where to start if you've never held a federal award

The barrier is rarely the technology. It's the absence of past performance and the absence of a national laboratory relationship — and DOE has an inexpensive fix for both.

The High-Performance Computing for Energy Innovation program awards up to $400,000 to small and medium-sized manufacturers paired with a national laboratory, university, or nonprofit, with more than $10 million available across selections. A solicitation went out this spring. These projects put your name on a DOE award, pair you with lab staff who become credible partners and letter-writers later, and generate the modeling and validation data that strengthens a subsequent multi-million dollar application.

It is the best-value on-ramp in the federal industrial portfolio and it is chronically under-applied to.

Three habits separate companies with federal pipelines from companies with federal aspirations. Respond to Notices of Intent and Requests for Information — DOE routinely publishes an NOI months ahead, which is your window to shape scope and assemble a team rather than scramble. Complete SAM.gov and application portal registrations before you need them, not during a three-week concept paper window. And line up cost share early: these are cooperative agreements, in-kind contribution from a host site frequently counts, and that host commitment letter is worth more than another month of narrative polish.

The through-line

Twenty projects. Fifteen states. Four universities working alone on membranes and materials, startups anchored to Tyson Foods and Rio Tinto, a national lab partnered with an Iowa ethanol plant, and a water utility and a tribal nation sitting quietly on partner lists next to Cleveland-Cliffs.

None of them won by guessing the right political vocabulary. They won with a quantified improvement over a real incumbent process, a team sized correctly to what they were promising, a credible answer to how the thing gets manufactured, and a budget that funded the validation work instead of hoping to absorb it.

That's repeatable. It's also, for most organizations, about nine months of preparation before the solicitation you actually want ever drops.

The organizations that will win the next tranche are assembling their teams right now.

Kristin Cooper is CEO and Founder of Grant Management Associates, which develops and manages federal grant strategy for clients across water, energy, transportation, broadband, defense, and tribal sectors.