The Best Treatment for Meth and Opioids Is a Gift Card. Silicon Valley Would Rather Sell You a Watch.

Bernard Groves had spent five years trying to quit methamphetamine — losing his job, his car, nearly his apartment — when a counselor at HealthRIGHT 360 in San Francisco offered him something stranger than another round of therapy: a Walmart gift card, starting at $10, every time his urine test came back negative. Twice a week, for months. “Like, how could you say you’re excited to pee in a cup? But I was, every week,” Groves told Tradeoffs in 2024, describing the ritual that got him through the longest stretch without meth he’d had in years.

The treatment with the strongest evidence base in addiction medicine is a gift card.

Nobody in venture capital is trying to build a company around that.

On September 18, 2026, the Substance Abuse and Mental Health Services Administration announced $42.3 million in new supplemental funding to states and territories, plus $2 million to the Addiction Technology Transfer Center Network, to push a “Treatment First” approach to homelessness and addiction. It’s real money, and it’s aimed mostly at workforce training, housing infrastructure, and getting agencies to talk to each other — worthwhile, unglamorous, and not really a technology story. The technology story is the one sitting next to it, mostly uncovered: eighteen months earlier, SAMHSA quietly multiplied by ten the amount a treatment program can legally pay someone for staying off drugs, from $75 a year to $750. That’s the intervention researchers call contingency management — plain terms, you get a small reward, right away, every time a drug test comes back negative, instead of waiting for a chip at a meeting or a certificate at 90 days. It is, by a wide margin, the best-studied behavioral treatment for stimulant use disorder that exists, and until last year it was capped at an amount too small to matter.

Meanwhile, the money that’s supposed to fix addiction treatment in this country is flowing somewhere else entirely: into wearables that track your heart rate and sleep to guess when you might use again, AI chatbots that text you when you seem stressed, and a wave of “recovery tech” venture funding that treats contingency management’s own hard-won evidence as background noise. This is a story about which one actually works, and about who gets to be trusted with $10 versus who gets watched.

Washington Finally Paid for the Cheapest Thing That Works

Contingency management isn’t complicated, which may be part of why it took the federal government three decades to fully embrace it. You get tested for the drug you’re trying to quit — usually urine, sometimes breath — on a regular schedule. Every negative result earns a reward, typically a retail gift card, and the value climbs the longer your streak runs. A positive test resets it to zero. That’s it. No pill, no app, no algorithm — a person handing another person a card because they didn’t use today.

Rick Rawson, a UCLA professor emeritus of psychology who has pushed for this treatment since the 1990s, told Tradeoffs that for years the addiction treatment field dismissed it outright. “You would hear things like, this isn’t treatment, this is just paying people not to use drugs,” Rawson said. “It wasn’t a medicine. It wasn’t talk therapy. It was this sort of transactional thing.” A California pilot Rawson helped design in 2005 got shut down by federal regulators over Medicaid fraud concerns before it ever proved itself. The treatment sat mostly unused for a decade, funded almost nowhere outside the VA.

What finally moved it wasn’t a new study — the evidence had been sitting there for years. A systematic review and meta-analysis covering 74 randomized controlled trials and more than 10,000 patients on medication for opioid use disorder found contingency management beat standard care across nearly every outcome researchers tracked, including keeping people on their buprenorphine. What moved the policy was fentanyl. As overdose deaths involving meth and cocaine climbed — driven largely by fentanyl contaminating the stimulant supply, since there’s no buprenorphine or methadone equivalent for stimulants — states ran out of alternatives. Rawson noted that shift directly to Tradeoffs: people were “buying cocaine or methamphetamine and dying of fentanyl overdoses because fentanyl is mixed into the drug supply.”

What finally moved it wasn’t a new study — the evidence had been sitting there for years.

That gap between how well studied contingency management is and how little it was actually used is the whole story in miniature: the treatment that did meaningfully better than standard counseling — solidly enough that researchers’ standard measure of that gap, called Cohen’s d, landed at 0.58 for illicit opioid use and 0.70 for stimulant use, comfortably in the medium-to-large range statisticians use to describe a real, reliable effect — sat capped at $75 a year for decades while nobody outside a research department paid attention.

California became the first state to fund contingency management through Medicaid in 2023, under its CalAIM waiver. The design is deliberately modest: gift cards start at $10 per negative test and climb to $26.50, capping around $599 a year — well under SAMHSA’s old $75-per-visit ceiling stacked across a year, and far under research showing the treatment works even better above $1,000. By September 2024, nearly 4,000 people had enrolled, and UCLA researchers found at least 75% of submitted urine samples came back negative for stimulants, according to Tradeoffs’ reporting. California’s Medicaid director, Tyler Sadwith, framed the state’s caution as strategic patience: “We need to prove that this works and that this works well,” he said, describing the pressure of being the first state to try it. The Washington State Institute for Public Policy — a nonpartisan state research office, not an advocacy group — has independently classified contingency management as a cost-beneficial intervention for opioid use disorder patients on medication, on top of the clinical outcomes.

It’s also not universally available, and that gap has a body count. Ayesha Appa, an addiction specialist who runs an HIV clinic at San Francisco General Hospital, offered contingency management through a private grant until it ran dry, and her hospital isn’t currently eligible to offer it through CalAIM. She told Tradeoffs about a 45-year-old patient who “desperately wanted to stop using” meth but didn’t trust the referral to a different clinic that could offer the incentive program. The woman died of an overdose. “When I think about her, it’s an equal mix of guilt and regret because it truly felt like we could have done more,” Appa said. That’s the stakes of a $750 cap increase that most people reading this have never heard of: not an abstraction, a specific and preventable gap between what works and who can get it.

The Money Went Looking for Something It Could Put on a Dashboard

While the federal government was raising a gift-card cap almost nobody covered, a very different kind of money was moving through addiction and behavioral health. Digital health investment hit roughly $4 billion in the first quarter of 2026 alone, with AI treated as a baseline expectation rather than a differentiator, per Fierce Healthcare’s tracking. Recovery-specific tech has ridden that wave: Huml Health sells addiction treatment centers a platform built around Samsung watches that collect biometric data — heart rate variability, sleep, movement — for clinicians and families to monitor. Empathy Health Technologies raised $7.6 million in December 2025 for its Sober Sidekick app, which uses what it calls an “Empathy Algorithm” to analyze behavioral patterns and flag relapse risk, citing an outside analysis claiming a 68% reduction in relapse risk for users who engage with the app just five times — a figure worth reading carefully, since it comes from a commissioned analysis by a firm the company itself cites in its marketing, not an independently replicated clinical trial.

There’s also a genuine category here worth taking seriously: an app that a doctor prescribes like a drug, reviewed and cleared by the FDA to treat a specific condition — the industry calls these prescription digital therapeutics. reSET-O, cleared by the FDA in 2018, is one of these: a 12-week app-delivered course of cognitive behavioral therapy for opioid use disorder, prescribed alongside buprenorphine. It’s a real, regulated, evidence-reviewed product — and there’s a telling wrinkle in its own origin story: the pivotal NIDA-sponsored trial that won it clearance measured its benefit in patients who were receiving buprenorphine plus contingency management, not the app in isolation. Even the government-cleared digital therapeutic built its case on contingency management’s shoulders. What’s changed since 2018 is that the industry around it has stopped citing that shoulder and started selling the app as the innovation.

Even the government-cleared digital therapeutic built its case on contingency management’s shoulders.

You’ve probably been pitched one of these, or something like it, if you’ve touched a treatment program in the last two years — a wearable “for accountability,” an app that promises to catch you before you slip. The pitch is seductive because it sounds like it’s doing something contingency management doesn’t: watching you when no counselor can. That’s also exactly the problem.

The Evidence Gap Nobody’s Advertising

Here’s the part the funding rounds don’t put in the press release: relapse-prediction wearables do not have anything close to contingency management’s evidence base, and the people doing the actual research say so plainly. A scoping review in Current Addiction Reports examining wearable biosensors in the substance use field found the field still working through basic feasibility questions — small samples, short observation windows, inconsistent accuracy across devices, and results that often don’t transfer from a lab setting to someone’s actual apartment. That’s a different category of evidence entirely from a 74-trial meta-analysis spanning 10,000-plus patients. One is a mature, replicated clinical literature. The other is, charitably, promising early-stage research being sold to venture capital and treatment centers as though it were already proven.

This isn’t a case for dismissing the technology outright — some of it may eventually earn its evidence. It’s a case against the current mismatch between confidence and proof. Wearables and AI chatbots get press coverage, investor decks, and glossy claims about “predicting relapse before it happens.” Contingency management, which the research has backed for over thirty years, got a $75 federal ceiling until fourteen months ago and still fights for state Medicaid slots one waiver application at a time. If you’re the kind of reader who’s sat through an intake assessment where a staffer handed you a tablet instead of looking you in the eye, you already know which of these two things felt like it was actually paying attention to you and which one felt like a product being tested on you.

This Is Also an Old Fight, Dressed in New Hardware

Addiction medicine has been here before. When the federal government first regulated methadone clinics in December 1972, it built the system around surveillance as a condition of trust: mandatory urine testing on a fixed schedule, medication dispensed under direct clinic observation, and take-home doses granted only after a patient demonstrated — through documented compliance, not their own account — that they’d earned the privilege, according to the National Academies’ history of methadone regulation. The logic was that people who use drugs can’t be trusted with autonomy, so autonomy has to be extracted one verified negative test at a time. That framework shaped fifty years of methadone-clinic culture, and it’s the same instinct that later fueled the harm reduction movement’s fight against drug testing as a gatekeeping tool rather than a health measure.

That instinct didn’t retire — it got a Bluetooth chip. In 2023, Filter’s Helen Redmond reported on a wave of companies selling GPS-enabled, remotely-lockable methadone lockboxes and video-verification apps to opioid treatment programs, marketed as “harm reduction” tools even as they let clinic staff surveil and, in some cases, remotely restrict access to a patient’s own take-home medication. Aaron Ferguson of the National Urban Survivors Union put it bluntly: “If history is any indication of how technology gets used when it comes to people who use drugs, then we can expect bad consequences.” Swap the GPS lockbox for a smartwatch that flags your heart rate to a case manager, and the underlying logic is the same one from 1972: we’ll trust you once the device confirms it. Contingency management inverts that logic entirely — it pays you first, on the honor of a single test result, no ongoing feed of your biometrics required. That’s arguably the more radical idea in this whole story, and it’s the one getting the smaller budget.

Contingency management inverts that logic entirely — it pays you first, on the honor of a single test result, no ongoing feed of your biometrics required.

What Real Money Is Actually on the Table Right Now

None of this means the SAMHSA news from September is nothing — $42.3 million for housing and workforce infrastructure matters, and the $2 million ATTC Network award will genuinely help programs train staff. It’s also, honestly, not a technology story or a contingency management story; SAMHSA said explicitly the funding covers “technical assistance, training, policy development, and cross-system coordination rather than the direct provision of treatment services.” The money that actually rewards the intervention with the evidence behind it is the quieter one: that $750 cap, sitting there since January 2025, still underused because most state Medicaid plans and treatment programs haven’t updated their billing structures to reflect it.

If you work at or run a program that already offers contingency management, the concrete thing to do this week is check whether your state’s Medicaid contingency management benefit — or your grant-funded protocol — has actually been updated past the old $75 ceiling; in a lot of states, the paperwork hasn’t caught up to the policy, and that gap is costing patients real money, the kind Bernard Groves used to buy weights, food for his pet bird, and a movie night with his mom, sister, and grandmother. This coverage sits alongside our earlier reporting on contingency management’s mechanics and fits the broader pattern we track in Rize’s technology and innovation coverage of opioid and fentanyl treatment: the tools that get funded aren’t always the tools that work, and the tools that work best are often the ones too plain to pitch to a Series A.

Naloxone is still free at pharmacies nationwide and federally funded regardless of how any of this shakes out. The $750 cap is real money on the table, right now, for any program willing to fill out the paperwork. The tech industry will keep pitching you a smarter way to watch yourself. The federal government just quietly agreed to pay for something dumber and better: a stranger, a cup to pee in, and a gift card that says someone noticed you tried.