ARPA-H's FASTPASS Wants to X-Ray Your Medicine Without Opening the Box — the Solicitation Drops in Mid-August

Every year, drug recalls in the United States reach more than 150 million Americans — and the defining feature of almost every recall is that it happens after the affected product has already been dispensed and, often, taken. By the time a substandard or contaminated medication is identified, the exposure has occurred. Current front-line defenses do not close that gap: FDA port inspections physically test only a small fraction of shipments and can take weeks to return results, and drug-related problems more broadly cost the U.S. healthcare system on the order of $528 billion a year.

ARPA-H's FASTPASS program is a bet that the answer is to check the medicine without ever opening it — to read a product's chemical makeup straight through its packaging, in minutes rather than weeks, at the scale of the supply chain rather than a sampled few percent. The program's solicitation is expected to post to SAM.gov in mid-August 2026, with a virtual Proposer Workshop on August 24 (registration required by August 20 at 12:00 p.m. ET, notice ID ARPA-H-SN-26-157) and a hybrid Proposers' Day to follow after publication. For teams that work in spectroscopy, pharmaceutical chemistry, sensing hardware, or machine learning, the window to get organized is now — not after the solicitation drops.

Here is what FASTPASS is asking for, and how to think about competing for it.

Two technical tracks: sense it, then act on it

FASTPASS is organized around two technical areas that map cleanly onto the two halves of the problem — detecting the defect, and turning that detection into a decision.

TA1 — SAMPLE. This track funds advanced through-packaging sensing technologies that can evaluate a product's chemical composition without breaking the seal. The emphasis is on spectroscopic methods — approaches that read the molecular signature of a substance through the bottle, blister pack, or shipping container. The hard part is not spectroscopy in a clean lab on an exposed sample; it is doing it through real-world packaging, with the interference, variability, and speed constraints that a working supply chain imposes. A winning TA1 concept has to survive contact with the messy physical reality of how drugs are actually packaged and moved.

TA2 — DETECT. Sensing raw signal is worthless if no one can interpret it fast enough to act. TA2 funds the analytics and models that transform detection techniques into practical, deployable tools — the software layer that converts a spectroscopic reading into a confident, actionable "this product is compromised" call. This is where machine learning does the work: distinguishing a genuine defect from packaging noise, calibrating confidence, and doing it at a speed and reliability level a regulator or distributor could actually rely on.

The two-track structure is a signal about how ARPA-H sees the problem. It is not funding a better spectrometer or a better algorithm in isolation; it is funding the pairing — a sensing modality and the intelligence to interpret it — because a detection capability that a human cannot act on at supply-chain speed does not solve the recall problem.

Why this is an ARPA-H problem, not an NIH one

It is worth understanding why a program like FASTPASS lives at ARPA-H rather than at a traditional health agency, because that shapes how you should approach it. ARPA-H exists to fund high-risk, high-reward health breakthroughs on aggressive, milestone-driven timelines — the DARPA model applied to health. It does not fund incremental science; it funds capabilities that do not yet exist and that, if they worked, would change the baseline.

Through-package chemical verification at supply-chain scale is exactly that kind of capability. No routine system today reads the contents of a sealed drug package in minutes across a meaningful share of the supply. Building one requires stitching together fields that rarely sit under one roof — spectroscopy, AI/ML, pharmaceutical chemistry, and sensing hardware engineering — which is why ARPA-H is explicitly encouraging cross-disciplinary teams and has stood up a teaming-profile page on the ARPA-H Solutions site to help potential partners find each other. The program manager is CAPT James Coburn.

For applicants, the practical implication is that a single-lab, single-discipline proposal is structurally disadvantaged. FASTPASS is asking for a system, and systems require teams.

The timeline is the strategy

The single most important thing to internalize about FASTPASS right now is the sequence of dates, because ARPA-H programs move fast once they open and the pre-solicitation window is where the real positioning happens.

• Mid-August 2026: the solicitation posts to SAM.gov. This is when the actual technical requirements, evaluation criteria, and submission mechanics become concrete.
• August 20, 12:00 p.m. ET: registration closes for the Proposer Workshop.
• August 24: the virtual Proposer Workshop — the single best opportunity to hear how the program office frames the problem and what it considers a compelling response.
• After publication: a hybrid Proposers' Day, and the opening of the response window.

The teams that compete well for ARPA-H programs treat the period before the solicitation as the decisive phase. That means: register for the August 24 workshop before the August 20 cutoff; use the pre-solicitation weeks to close the disciplinary gaps in your team (if you have the spectroscopy but not the ML, or the hardware but not the pharmaceutical-chemistry credibility, find the partner now via the teaming site); and be ready to move the moment the solicitation posts, because ARPA-H response windows are typically short and demand a crisp, milestone-structured proposal rather than an open-ended research plan.

How ARPA-H evaluates — and how to pitch

ARPA-H does not reward the most thorough literature review or the most exhaustive methods section. It rewards a credible, quantified path to a capability that does not exist yet, expressed in milestones and metrics. Three things distinguish a strong FASTPASS response.

A specific, measurable performance target. Vague promises to "improve drug screening" lose. ARPA-H thinks in numbers: what fraction of the supply can your approach cover, at what throughput, with what false-positive and false-negative rates, through what range of real packaging? A proposal that commits to hard metrics — and lays out the risk-reduction path to hitting them — reads as ARPA-H-native.

Honest engagement with the hard part. For TA1, that is almost always the packaging interference problem; for TA2, it is confidence calibration and false-alarm rates at operational speed. Naming the genuine technical obstacle and showing a defensible plan to overcome it is far more persuasive than pretending it is easy.

A path to the real world. Because the goal is a deployable capability, the strongest teams can articulate how the technology would actually integrate into the supply chain — port inspection, distribution, pharmacy — and, where relevant, how it maps to the regulatory reality that governs drug testing. A brilliant sensor with no plausible deployment story is a science project, not a FASTPASS answer.

The bottom line

FASTPASS is aiming at a problem that current systems structurally cannot solve: catching bad medicine before patients take it, at a scale and speed the existing sampled, weeks-long inspection regime never will. For teams with the right mix of spectroscopy, sensing hardware, pharmaceutical chemistry, and machine learning, it is a rare chance to build a genuinely new public-health capability with ARPA-H's speed and budget behind it. The solicitation is days away. The move this week is to register for the August 24 workshop before the August 20 deadline, and to spend the pre-solicitation window assembling the cross-disciplinary team a program like this demands — because once the solicitation posts, the clock will already be running.