AI, Enzymes, and the Factory Floor: A Federal Test for Plastic Circularity

BERKELEY, Calif. – October 14, 2026 — Industrial progress is rarely defined by a single, sudden breakthrough in a pristine laboratory. More often, it is forged through the slow, complex integration of a radical new idea into an entrenched legacy system. For decades, the global packaging and plastics industry has relied on mechanical shredding and melting to recycle polyethylene terephthalate (PET). It is a brute-force approach that works reasonably well for clean water bottles, but it fails spectacularly when faced with contaminated food trays, mixed-material packaging, or synthetic polyester textiles. Now, the U.S. Department of Energy (DOE) is stepping in to determine if artificial intelligence and biotechnology can bridge that critical gap on a high-speed factory floor.

In an announcement that merges advanced bio-engineering with heavy industrial manufacturing, biotechnology startup Rhea’s Factory and beverage manufacturer Niagara Bottling have been selected for the DOE’s Industrial Technology Validation (ITV) program. The mandate is straightforward but daunting: install and operate a proprietary enzymatic recycling platform, dubbed RheaCycle, under grueling, real-world production conditions at Niagara’s PET recycling facility in Vernon, California.

Administered by the DOE’s Industrial Technologies Office, the ITV program recently distributed a combined $3 million in funding across nine nationwide projects. The initiative pairs technology developers with industrial host sites to objectively assess emerging green technologies. For the plastics industry, which has long struggled with the economic and chemical limitations of traditional recycling, this pilot represents a critical stress test. It is an opportunity to prove whether bio-recycling can move beyond the realm of startup hype and become a viable, scalable pillar of modern manufacturing infrastructure.

Biology Meets the Bottler

To understand the significance of this deployment, one must look at the physical realities of the host site. In May 2026, Niagara acquired the Vernon recycling and packaging facility, along with the rPlanet Earth brand, to secure a domestic supply of high-quality recycled PET (rPET) flakes and pellets for its own packaging operations. The facility is a massive, high-throughput environment designed to feed one of the largest family-owned beverage manufacturers in the world.

Integrating a biological process into this mechanical ecosystem requires a hybrid approach. The project will install the RheaCycle system as a parallel line at the Vernon site. This allows the beverage giant to continue operating its highly efficient mechanical recycling lines for clean bottle streams without interruption. Meanwhile, the enzymatic line will act as an industrial safety net, catching and processing the complex, off-spec feedstocks that the mechanical sorters reject.

The science behind the new line is fundamentally different from traditional recycling. Instead of melting the plastic—which degrades its molecular integrity over successive cycles—the bio-platform utilizes AI-engineered enzymes to execute a targeted chemical breakdown. Operating at low temperatures and atmospheric pressure, these biocatalysts selectively depolymerize PET into its two foundational building blocks: purified terephthalic acid (TPA) and ethylene glycol (EG).

Because the enzymes act only on the PET polymer, the system can digest heavily contaminated inputs, including thermoformed fruit containers, residual streams from existing recycling lines, and even polyester textiles. The resulting recovered monomers are virgin-equivalent. They can be repolymerized into new, food-grade PET infinitely, without any loss of quality. By running these two technologies side-by-side, the Vernon plant is pioneering a model where mechanical and biological systems complement each other, maximizing the total volume of plastic diverted from landfills.

The Federal Litmus Test

The history of clean technology is littered with promising startups that achieved miraculous results in a beaker, only to face bankruptcy when attempting to scale those reactions to thousands of gallons in a continuous industrial process. This scaling “valley of death” is precisely what the DOE’s ITV program is designed to navigate.

Rather than relying on the optimistic projections of the technology developer, the federal government is deploying experts from Lawrence Berkeley National Laboratory (LBNL) to conduct independent measurement and verification. The national lab will meticulously track the system's performance, focusing heavily on resource efficiency. Energy and water consumption per kilogram of recovered monomer will be independently measured, verified, and eventually published.

This level of transparency is rare but necessary. By providing objective, third-party data on hard metrics like kilowatt-hours and liters of water utilized, the DOE aims to de-risk the technology for future investors and industrial adopters.

“Bringing a technology like this to commercial scale takes more than good science,” said Dr. Arzu Sandikci, CEO and Co-founder of Rhea’s Factory. “It takes a federal program willing to fund independent validation, a national laboratory willing to measure it, an industry partner like Niagara willing to host it, and a startup that has already proven the science and built the process to run at industrial scale. This project brings all four together, and that’s how a new industry gets built.”

Dr. Sandikci’s perspective is grounded in the broader federal push for industrial decarbonization. As a current innovator in the West Gate program at the U.S. Department of Energy’s National Laboratory of the Rockies, she and her team have been actively working to optimize downstream monomer purification and drive down operational costs. The Vernon pilot is the crucible where those optimizations will be tested against the relentless pace of a commercial bottling supply chain.

AI Enzymes vs. Fossil Polymers

Ultimately, the success of this technology will not be judged on its environmental benefits alone, but on its economic viability. The global plastics market is a commodity business, heavily tethered to the fluctuating price of crude oil. For enzymatic recycling to achieve true circularity at scale, the cost of the recovered TPA and EG monomers must reach parity with virgin, petroleum-derived PET resin.

This economic imperative is where the Berkeley-based startup's specific technological approach becomes critical. Traditional chemical recycling methods, such as pyrolysis or gasification, often require massive energy inputs to generate the extreme heat and pressure necessary to break molecular bonds. By contrast, enzymatic depolymerization leverages biology to do the heavy lifting at near-ambient conditions.

However, natural enzymes are notoriously fragile and slow. To overcome this, the company utilizes advanced machine learning models to design and engineer highly thermostable, hyper-efficient biocatalysts in a matter of months rather than years. This AI-driven approach to protein engineering is designed to accelerate the catalytic speed and extend the lifecycle of the enzymes, directly driving down the per-unit cost of the recovered monomers.

The potential applications for this technology extend far beyond beverage packaging. Earlier this year, the biotechnology firm was recognized with a major global award for utilizing its AI-designed enzymes to enable textile-to-textile recycling, addressing the massive global challenge of synthetic fashion waste. But for now, the focus remains squarely on the rigid packaging sector and the immediate demands of the commercial pilot.

Site preparation is currently underway in Southern California. The complex process of installing and commissioning the parallel biological line is expected to take up to six months before the formal validation period begins and the federal sensors start recording data.

If the metrics align with the laboratory models, the implications for the manufacturing sector will be profound. It will signal a shift away from a purely extractive industrial economy toward one that treats its own engineered waste as a high-value, infinitely renewable feedstock. The data generated in Vernon over the coming year will likely determine whether the future of plastic circularity belongs to the brute force of the past, or the engineered biology of the future.