Trend analysis: Waste sorting & recycling robotics — market signals, investment flows, and the 2026–2028 outlook

Why It Matters

The global waste sorting robotics market surpassed US $560 million in 2024 and is on track to exceed US $1.3 billion by 2028, growing at a compound annual rate of roughly 24 percent (Grand View Research, 2025). Behind that headline number sits a convergence of forces that makes the 2026 to 2028 window uniquely consequential: the European Union's packaging and packaging waste regulation (PPWR) mandates recycled-content floors of 25 percent for PET bottles by 2025 and 30 percent for all plastic packaging by 2030, while extended producer responsibility (EPR) programmes in North America are expanding rapidly, with five US states and three Canadian provinces launching or strengthening EPR frameworks between 2024 and 2026 (OECD, 2025). These regulatory shifts create binding demand for cleaner, higher-purity recyclate, and robotic sorting is emerging as the technology most capable of bridging the gap between current MRF output quality and the recycled-content specifications that brand owners and resin buyers now require. Understanding where the market is heading, which technology bets are paying off, and where capital is concentrating helps sustainability professionals, investors, and municipal planners make better procurement and allocation decisions over the next three years.

Key Concepts

Multi-material AI recognition. Early robotic sorters classified materials into broad categories such as PET, HDPE, and fibre. The current generation of AI vision platforms, led by AMP Robotics' Cortex and Machinex's SamurAI, can distinguish dozens of sub-categories including food-grade vs. non-food-grade PET, coloured vs. natural HDPE, and specific polymer resin codes within flexible film streams. AMP Robotics reported that its network-wide model can now identify over 100 distinct material classes with a weighted accuracy of 92 percent, up from 85 percent in 2023 (AMP Robotics, 2025).

Soft-grip and adaptive end effectors. Traditional suction cup grippers struggle with wet, flat, or irregularly shaped items. Companies such as Soft Robotics (acquired by Fanuc parent entities in 2024) and ZenRobotics (now part of Terex) have introduced pneumatic finger grippers and hybrid suction-pinch designs that expand the range of objects a single robot can handle. Field data from Machinex installations show that adaptive grippers increase successful pick rates on flexible packaging by 15 to 20 percent compared to suction-only designs (Machinex, 2025).

Digital material passports and EPR data integration. As EPR schemes require producers to report the recyclability and recycled content of their packaging, MRFs need granular composition data. Robotic sorting platforms now generate item-level classification logs that feed directly into EPR compliance reporting, creating a data feedback loop between collection, sorting, and producer responsibility obligations (OECD, 2025).

Chemical recycling feedstock preparation. Advanced (chemical) recycling processes such as pyrolysis, solvolysis, and depolymerisation require tightly specified feedstock with low contamination. Robotic sorting is increasingly positioned as the technology that prepares these feedstock streams, extracting target polymers at purities of 95 percent or higher from mixed plastic bales (Closed Loop Partners, 2025).

Signals to Watch

Signal 1: Venture and growth-equity investment is accelerating. AMP Robotics raised a US $91 million Series D in late 2024, bringing its total funding to over US $250 million (Crunchbase, 2025). EverestLabs, which provides AI-overlay software for existing optical sorters, closed a US $16.5 million Series A in 2024 and announced expansion into European markets. Greyparrot, a London-based waste analytics platform, raised US $10 million in 2025 to scale its camera-based waste composition monitoring for MRFs and incineration facilities (Greyparrot, 2025). Investment is also flowing from strategic acquirers: Terex completed its acquisition of ZenRobotics in 2023, and Machinex has expanded its robotics division with dedicated AI research labs in Plessisville, Quebec. The trajectory signals that the industry is moving from early-stage experimentation toward commercial scale-up.

Signal 2: EPR mandates are creating binding demand for higher purity. Oregon, Colorado, Minnesota, California, and Illinois now have active EPR legislation that shifts the cost of packaging waste management to producers, who in turn demand higher-quality recyclate from MRFs to meet their recycled-content targets. In Europe, the PPWR's mandatory design-for-recycling criteria will phase out packaging formats that cannot be sorted at scale, effectively standardising the input stream for robotic systems and improving their effectiveness (European Commission, 2025). These regulatory tailwinds are structural, not cyclical, giving MRF operators confidence to invest in automation with multi-year payback horizons.

Signal 3: Robotics-as-a-Service (RaaS) models are lowering adoption barriers. AMP Robotics and Machinex both offer pay-per-pick or monthly subscription pricing that eliminates the upfront capital expenditure barrier for smaller MRFs. AMP's RaaS contracts now cover more than 40 percent of its new installations, according to company disclosures (AMP Robotics, 2025). This shift mirrors the broader enterprise software trend toward SaaS and allows operators to align costs with throughput, reducing financial risk during commodity price downturns.

Signal 4: AI-powered waste composition analytics are becoming a standalone product category. Greyparrot and Recycleye have deployed camera systems at MRF intake points, on conveyor belts, and at residual-waste transfer stations to generate real-time composition data. These analytics platforms do not sort material themselves but provide the data layer that optimises robotic and optical sorting downstream. SUEZ deployed Greyparrot's system across 19 UK and European facilities in 2024 and 2025, using composition data to adjust sorting parameters dynamically and improve fibre recovery by 4 percentage points (Greyparrot, 2025).

Signal 5: Construction and demolition (C&D) waste is the next frontier. Most robotic sorting deployments to date have targeted municipal single-stream recycling. However, C&D waste represents a US $150 billion global market with sorting needs that are poorly served by manual labour alone. Recycleye deployed its first C&D sorting system in the UK in late 2025, targeting wood, aggregates, and metals on a mixed C&D line. ZenRobotics has a longer track record in C&D sorting in the Nordic countries, and the segment is expected to account for 15 to 20 percent of new robotic sorting installations by 2028 (Lux Research, 2025).

Where the Value Pools Are

High-purity PET and HDPE recyclate. Food-grade recycled PET (rPET) commands a US $200 to $400 per tonne premium over virgin PET in European markets as of early 2026, driven by PPWR recycled-content mandates and voluntary brand commitments from Nestlé, Unilever, and PepsiCo (ICIS, 2026). Robotic sorting enables MRFs to produce bales with purity levels that meet food-contact standards, unlocking this premium. The value pool here is direct and measurable: a mid-sized MRF processing 30,000 tonnes of PET per year that improves bale purity from 92 to 98 percent can capture an additional US $300,000 to $600,000 in annual commodity revenue.

Chemical recycling feedstock. Closed Loop Partners estimates that the market for sorted, specification-grade feedstock for chemical recycling will reach US $2.5 billion globally by 2028, up from under US $400 million in 2024 (Closed Loop Partners, 2025). Robotic and AI-enhanced optical sorting is the primary technology for producing these feedstocks from mixed plastic waste that would otherwise go to landfill or incineration. Companies such as Plastic Energy, PureCycle Technologies, and Eastman are signing long-term feedstock offtake agreements with MRFs that can deliver consistent quality.

Data and compliance services. As EPR regulations expand, the data generated by AI sorting systems becomes a revenue stream in its own right. MRFs that can provide auditable composition reports, chain-of-custody documentation, and real-time contamination alerts to producer responsibility organisations (PROs) and brand owners can charge data fees or secure preferred supplier agreements. Greyparrot has built a business model around selling waste composition data to municipalities, PROs, and FMCG companies, demonstrating that the data layer can be monetised independently of the physical sorting (Greyparrot, 2025).

RaaS and platform economics. Companies that aggregate sorting data across hundreds of installations gain compounding advantages: larger training datasets improve AI accuracy, which attracts more customers, which generates more data. AMP Robotics' installed base of over 250 facilities creates a network effect that is difficult for later entrants to replicate. The platform economics of the robotics layer increasingly resemble those of enterprise SaaS, with high gross margins on software licensing and recurring revenue streams that stabilise during commodity price volatility.

Retrofit and upgrade market. An estimated 5,000 to 7,000 MRFs operate globally, and the vast majority were built before AI-driven robotics became commercially available. The retrofit opportunity, adding robotic cells and AI analytics to existing lines, is larger than the greenfield new-build market and carries lower integration risk because operators can start with a single cell and scale incrementally. EverestLabs has positioned itself squarely in this segment by offering software upgrades for existing optical sorters, capturing value without requiring full equipment replacement (EverestLabs, 2025).

Action Checklist

• Map regulatory exposure. Identify which EPR and recycled-content mandates apply to your facility's output markets and timeline the compliance milestones that will drive demand for higher purity.
• Evaluate RaaS vs. CapEx models. Compare total cost of ownership under purchase, lease, and pay-per-pick structures across a range of throughput and commodity price scenarios.
• Pilot on your highest-value stream. Start robotic sorting on PET or aluminium lines where commodity premiums justify the investment and where purity improvements translate directly to revenue.
• Invest in data infrastructure. Ensure your MRF's IT systems can ingest, store, and export the classification data that robotic and analytics platforms generate. This data is a future revenue source.
• Engage chemical recyclers early. If your facility handles mixed plastic fractions that cannot be mechanically recycled to food-grade, explore feedstock offtake agreements with advanced recyclers who need specification-grade input.
• Plan workforce transition. Retrain manual sorters for robot oversight, maintenance, and data quality roles. Communicate the transition plan to unions and workforce development partners proactively.
• Track the C&D opportunity. If your operations include C&D waste, monitor robotic sorting pilots in this segment and budget for potential deployment in the 2027 to 2028 timeframe.

FAQ

How does AI waste recognition differ from traditional optical sorting? Traditional NIR optical sorters identify materials by their spectral signature and use air jets to divert them. AI-enhanced systems add a visual recognition layer that classifies objects by shape, colour, label, brand, and context before activating a robotic arm or air jet. This means AI systems can distinguish between, for example, a food-grade PET bottle and a non-food-grade PET clamshell that have identical spectral signatures but different market values. The two technologies are complementary: optical sorters handle bulk separation at high speed, while robots perform targeted positive picks and quality-control sorting.

What is the typical payback period for a robotic sorting installation in 2026? Payback varies significantly by geography, labour cost, commodity mix, and financing structure. In North America and Western Europe, operators report 18 to 30 months for high-throughput PET and aluminium applications under CapEx purchase models. RaaS contracts eliminate the upfront capital but carry higher per-unit costs over a five-year horizon. In regions with lower labour costs, payback can extend to 36 to 48 months. Operators should model payback under conservative commodity price assumptions rather than relying on vendor base cases that often assume peak pricing.

Will robotic sorting eliminate the need for deposit return schemes (DRS)? No. DRS and robotic sorting serve different functions. DRS systems incentivise consumer behaviour, achieve collection rates of 85 to 95 percent for covered containers, and produce a clean, low-contamination input stream. Robotic sorting improves the quality of material that arrives at a MRF, regardless of how it was collected. The two systems are complementary: countries with both DRS and advanced MRF automation, such as Norway and the Netherlands, consistently achieve the highest recycling rates in Europe (Reloop Platform, 2025).

Are there enough skilled technicians to maintain robotic sorting systems at scale? Workforce availability is a genuine constraint. Robotic sorting systems require technicians with skills in industrial automation, PLC programming, and basic data science. AMP Robotics and Machinex have both launched certified technician training programmes, and several US community colleges now offer two-year degrees in recycling automation technology. However, the installed base is growing faster than the trained workforce, and maintenance response times in rural areas can be problematic. Remote diagnostics and over-the-air software updates mitigate some of this risk but cannot replace on-site mechanical intervention when hardware fails.

How will AI sorting interact with packaging design-for-recycling standards? The relationship is synergistic. As the PPWR and similar regulations mandate design-for-recycling criteria, packaging will become more standardised and easier for AI systems to classify. At the same time, the composition data generated by AI sorting platforms feeds back to producers, showing them which packaging formats are successfully sorted and which are not. This data loop is already active: Greyparrot provides brand-level sortability reports to FMCG clients, enabling them to redesign problematic formats before regulatory deadlines take effect (Greyparrot, 2025).

Sources

• Grand View Research. (2025). Waste Sorting Robotics Market Size, Share & Trends Analysis Report, 2024-2028.
• OECD. (2025). Extended Producer Responsibility and the Transition to a Circular Economy: Updated Policy Guidance.
• AMP Robotics. (2025). Cortex Platform: Network Performance, Material Classification, and RaaS Deployment Data.
• Machinex. (2025). SamurAI Sorting System: Adaptive Gripper Performance and Multi-Material Recognition Benchmarks.
• EverestLabs. (2025). AI Overlay for Optical Sorters: Retrofit Economics and European Market Expansion.
• Greyparrot. (2025). Waste Composition Analytics: Deployment Results Across SUEZ Facilities and Brand Sortability Reporting.
• Closed Loop Partners. (2025). Bridging the Gap: Feedstock Quality Requirements for Chemical Recycling and the Role of Advanced Sorting.
• European Commission. (2025). Packaging and Packaging Waste Regulation: Design-for-Recycling Criteria and Recycled-Content Targets.
• Crunchbase. (2025). AMP Robotics Funding History and Investor Profile.
• ICIS. (2026). European Recycled Polymer Pricing: rPET and rHDPE Premium Analysis, Q1 2026.
• Lux Research. (2025). Robotic Sorting for Construction & Demolition Waste: Market Sizing and Technology Readiness.
• Reloop Platform. (2025). Global Deposit Return Systems: Collection Rates, Material Quality, and System Design Comparison.