☰

UPSIDE Foods

Prove out scale-up without burning batches

Industry
Cultivated Meat and Cell Culture
Headquarters
Berkeley, California, United States
Public information as of
March 2026

A4BEE prepared this analysis from publicly available sources. It reflects our own reading of UPSIDE Foods's published strategy and is not endorsed by, or produced in cooperation with, UPSIDE Foods. Company website

Strategic priorities

UPSIDE Foods is the most heavily capitalized company in cultivated meat, having raised roughly $600 million across Series A through C, with a Series C of $400 million in April 2022 predicated on commercial-scale production. Its near-term proof point now sits entirely at the EPIC facility in Emeryville, California — a retrofitted 53,000-square-foot building being expanded from 50,000 to a theoretical 400,000 pounds of annual capacity — after the $141 million Rubicon greenfield plant in Glenview, Illinois was paused in February 2024 to conserve capital.

The central technical question is scale. Management has said it has run dozens of batches at the 2,000-liter scale, and the strategy rests on turning those runs into a repeatable process. Bioprocess scale-up is non-linear: increasing vessel volume changes mixing time, gas dispersion and nutrient availability, so parameters proven at 2 liters do not carry to 2,000 without re-work. The company describes bridging that gap through physical runs, where each 2,000-liter batch commits a significant amount of cell culture media.

Two parallel priorities lean on the same digital foundation. In January 2026 UPSIDE spun out Lucius Labs, a B2B life-sciences business selling cell culture media, buffers and analytical services covering 63 targets — a pivot that needs pharmaceutical-grade quality systems and customer-facing infrastructure of its own. And its ground-textured suspension products, which are easier to scale than whole-cut tissue, are moving through the FDA and USDA dual-oversight path that governs everything the company sells. All three — scale-up, Lucius Labs, and regulatory clearance — rest on process and quality data that can be read outside the equipment that produced it.

Challenges we see

  • Operations Manufacturing

    Carrying process behaviour from 2 litres to 2,000 litres

    UPSIDE's value proposition depends on scaling cell culture from small development vessels to 2,000-liter bioreactors. Fluid dynamics, shear stress and oxygen transfer all behave differently at volume, so a recipe that holds at bench scale has to be re-established at pilot scale. Management has said it has run dozens of batches at 2,000 liters.

    Where the scale-up path is walked through physical batches, each run commits media and time to learn one point in the parameter space. Modelling the vessel first narrows the number of physical runs to the ones that test a specific hypothesis, so the same budget buys more learning.

  • Digital Integration

    Letting biological signals reach the bioreactor's controls

    Biological data from R&D experiments and the PLC and SCADA systems governing the pilot bioreactors sit in separate estates. Mechanical set-points — agitation, sparging, temperature — are set independently of the live biological picture, and the research notes cell-viability losses when those set-points do not track cell conditions.

    When the control layer cannot read cell-viability signals, it holds set-points that were correct for an earlier state of the culture. Closing that loop lets agitation, sparging and temperature respond to the culture as it actually is, rather than to the plan it was started on.

  • Operations Manufacturing

    Raising volumetric output per operator on the whole-cut line

    The FDA/USDA-cleared whole-cut chicken product is made in manually operated 2-liter roller bottles, each yielding a few grams of tissue against roughly 170 grams for a conventional chicken breast, with scientific staff handling every step.

    At that yield-per-vessel, output scales with skilled labour rather than with equipment, which is what keeps whole-cut unit costs high. Automating rotation, media exchange and harvest shifts the constraint from operator hours to line throughput, and makes production volumes forecastable for retail and foodservice partners.

  • Compliance Regulatory

    Keeping one auditable thread across FDA and USDA oversight

    UPSIDE operates under a split where the FDA covers cell collection, banking and cultivation and the USDA covers harvest, processing and labelling, so a single product needs one continuous record from biopsy to package that satisfies two agencies with different inspection standards.

    When batch genealogy is assembled from separate lab and quality systems for each review, compiling and checking it becomes the work rather than the technical judgement inside it. A continuous digital thread makes real-time batch release and root-cause analysis a property of the record instead of a reconstruction.

  • Digital Operations

    Standing up Lucius Labs' quality and customer systems

    Lucius Labs launched in January 2026 as a B2B life-sciences vendor selling media, buffers and analytical services across 63 targets. That model needs secure customer portals, formulation software and a validated quality system able to satisfy pharmaceutical and biotech client audits — a different shape from food-production infrastructure.

    A B2B life-sciences customer expects certificates of analysis, audit trails and validated documentation as part of the product. Building that quality and customer-facing layer early is what lets Lucius Labs onboard clients and recognise revenue on the timeline the spinoff was created to hit.

Opportunities, by urgency and business impact

Each bubble is one opportunity, numbered to match the list below. Further right means it bites sooner; higher means a bigger effect on the business. A bigger bubble means a bigger implementation effort.

Source: A4BEE analysis of public sources

  1. Modelling the 2,000-litre vessel before the batch runs

    The gap between small development vessels and 2,000-liter tanks is currently bridged through physical runs, where each batch commits media and time to learn one set of parameters. Fluid dynamics, metabolic consumption and thermal gradients all shift with vessel volume.

    A digital twin of the bioreactor — computational fluid dynamics for mixing and sparging, plus a metabolic model of nutrient consumption — lets parameter combinations be tested virtually first, so the physical runs are the ones chosen to confirm a specific answer rather than to search for it.

    • UPSIDE Foods deep research profile, EPIC facility and 2,000-liter scale-up, March 2026
    • The company's COO, on translating the science at scale

  2. Closing the loop between cell signals and bioreactor set-points

    Biological sensor data and the PLC/SCADA controllers governing the bioreactors are structurally separate, so mechanical set-points do not track live cell viability and the research notes viability losses as a result.

    An OPC UA (Open Platform Communications Unified Architecture) layer streams pH, dissolved oxygen and viable cell density into the control system, so agitation, sparging and temperature can respond to the culture's real state instead of a fixed profile.

    • UPSIDE Foods deep research profile, IT/OT convergence, March 2026
    • The company's Chief Supply Chain & Manufacturing Officer, on manufacturing footprint

  3. Automating the whole-cut roller-bottle line

    Whole-cut production depends on manual roller-bottle handling by scientific staff, and suspension culture at scale needs tight, automated environmental control to hold yield steady and keep contamination risk down.

    Collaborative robots for rotation, media exchange and harvest, orchestrated by a manufacturing execution system that records batch genealogy and enforces recipe parameters, lets output scale with line throughput and gives partners a forecastable volume.

    • UPSIDE Foods deep research profile, manual production and suspension culture, March 2026

  4. Giving Lucius Labs a validated data and customer backbone

    As a new B2B life-sciences vendor, Lucius Labs needs media-formulation software, secure customer portals and a validated quality system before it can serve pharmaceutical clients and recognise revenue against the 63 analytical targets it offers.

    An integrated ELN-LIMS-QMS platform captures formulation development with full audit trails, links it to batch records, and exposes a customer portal for orders, status and certificates of analysis — the documentation a life-sciences buyer expects as part of the product.

    • UPSIDE Foods deep research profile, Lucius Labs launch January 2026, March 2026
    • The company's CTO, on reinventing media development

  5. Making batch genealogy readable across the FDA/USDA split

    Dual FDA/USDA oversight requires one continuous record from biopsy to package, but where lab, manufacturing and quality data sit in separate systems, compiling that genealogy for each review is a manual, error-prone step.

    An ontology-based platform harmonises SCADA, LIMS and ELN records into one queryable model, so quality data flows into submission-ready reporting and the digital thread is continuous rather than reassembled per inspection.

    • UPSIDE Foods deep research profile, FDA/USDA dual oversight, March 2026

What we'd propose

  • Digital CDMO

    Bioreactor digital twin for scale-up

    A virtual replica of the EPIC bioreactor environment that ingests historical process data and applies computational fluid dynamics with machine learning, so scaling parameters are explored and narrowed in simulation before a physical batch is committed.

    • CFD model of the vessel

      Mixing and sparging in simulation

      Build a computational fluid dynamics model of impeller-driven mixing, gas sparging and thermal gradients in the 2,000-liter vessel geometry, so the mechanical environment a culture will experience is known before the tank is filled.

    • Metabolic prediction layer

      Nutrient and oxygen demand curves

      Train models on historical batch data to predict nutrient consumption, oxygen demand and harvest windows for both tissue and suspension pathways, so feed and harvest decisions are informed by the culture's expected trajectory.

    • Virtual batch optimiser

      Test parameters before committing media

      Sweep agitation, sparging and temperature set-points in simulation to shortlist the configurations worth running physically, so each real 2,000-liter batch confirms a specific hypothesis rather than searching the space.

    • Physical runs are chosen to confirm an answer, so the media budget buys more learning per batch.
    • The mechanical environment at scale is understood before the tank is filled, not after the batch fails.
    • Scale-up evidence for the next stage is generated from a model the team can inspect and rerun.

    We've done this before

    See how we do Digital CDMO
  • Digital CDMO

    IT/OT convergence for closed-loop bioreactor control

    An OPC UA connectivity layer and edge-processing tier that stream biological sensor data into the PLC and SCADA control loop, so bioreactor set-points can respond to live cell conditions instead of a fixed profile.

    • OPC UA data bridge

      One protocol between lab and floor

      Stream pH, dissolved oxygen and viable cell density from biological sensors to the PLC and SCADA controllers through OPC UA (Open Platform Communications Unified Architecture), so process values leave the equipment in a documented, vendor-neutral form.

    • Edge processing at the vessel

      Local decisions, low latency

      Place edge nodes at each bioreactor for local preprocessing and anomaly detection, so latency-critical adjustments happen at the vessel while aggregated data still flows to the central platform.

    • Closed-loop control logic

      Set-points that track the culture

      Add adaptive control that modulates agitation, sparging and temperature against real-time viability signals, so set-points follow the culture's current state rather than the profile it was started on.

    • Biological signals and mechanical set-points sit in one loop instead of two separate estates.
    • Deviations are acted on at the vessel in real time rather than read from a later report.
    • The same connectivity model extends from the pilot bioreactors to future production lines.

    We've done this before

    See how we do Digital CDMO
  • Digital Lab

    Line automation and MES for cultivated meat production

    Collaborative-robot handling and a manufacturing execution system across the roller-bottle and suspension lines, so output scales with throughput, batch genealogy is recorded automatically and recipes are enforced with deviation alerts.

    • Cobot-driven flask handling

      Rotation, media exchange, harvest

      Integrate collaborative-robot stations for roller-bottle rotation, media exchange and harvest, so throughput per operator rises and the routine handling steps are performed to a consistent standard.

    • MES batch orchestration

      Genealogy from seed to product

      Deploy a manufacturing execution system to orchestrate the workflow, record complete batch genealogy from cell seed to finished product, and enforce recipe parameters with automated deviation alerts.

    • Yield analytics dashboard

      Yield and cost in view

      Surface yield, efficiency and resource-consumption metrics across lines, so cost-of-goods decisions are made against measured production data rather than estimates.

    • Output scales with line throughput rather than with skilled operator hours.
    • Batch genealogy is recorded as the line runs, so compliance evidence is a by-product of production.
    • Production volumes become forecastable for retail and foodservice partners.

    We've done this before

    See how we do Digital Lab
  • Digital Lab

    Lucius Labs digital platform and validated QMS

    An integrated ELN-LIMS-QMS backbone and B2B customer portal for Lucius Labs, so media formulation is captured with audit trails, linked to batch records, and exposed to clients as validated documentation.

    • Integrated ELN-LIMS

      Formulation captured as data

      Deploy a connected ELN-LIMS environment that captures formulation development, links it to production batch records and provides audit trails across the 63 analytical targets Lucius Labs offers.

    • Customer integration portal

      Orders, status, certificates

      Build a secure portal where clients submit custom formulation requests, track order status and retrieve certificates of analysis and validated quality documentation.

    • Validated QMS framework

      Audit-ready from day one

      Implement a GAMP5-aligned quality management system with electronic signatures, change control and CAPA tracking, so the business can meet audits from external life-sciences clients.

    • Lucius Labs can onboard clients with validated documentation from launch rather than building it under audit pressure.
    • Formulation work is captured as reusable data linked to the batch that used it.
    • Certificates and audit trails are part of the product, which is what a life-sciences buyer expects.

    We've done this before

    See how we do Digital Lab
  • Enterprise AI

    Ontology-driven data platform for FDA/USDA traceability

    An ontology-based platform that harmonises R&D, manufacturing and quality data into one queryable model, so the digital thread from biopsy to package is continuous and regulatory reporting draws from a single source.

    • Ontology-based data model

      One model across the sources

      Model the shared entities — batch, cell line, vessel, lot, result — once, then load SCADA, LIMS and ELN records against them, so a query returns comparable answers instead of reconciling several systems by hand.

    • Automated compliance reporting

      Submission-ready by default

      Compile batch genealogy, quality metrics and deviation records into FDA- and USDA-shaped reporting formats automatically, so review time goes on judgement rather than on transcription.

    • End-to-end digital thread

      Biopsy to package, traceable

      Track biological material from cell biopsy through cultivation, harvest, processing and packaging, satisfying both FDA cultivation traceability and USDA meat-inspection requirements from one record.

    • Batch genealogy is queried from one model rather than reassembled per inspection.
    • The digital thread is continuous across the FDA and USDA jurisdictional split.
    • New products and future sites attach to the model instead of triggering another integration.

    We've done this before

    See how we do Enterprise AI

Where QB Systems fits

Alongside our services we build QB Systems, hardware and software for bioprocess control. QB Systems is a product brand of A4BEE Sp. z o.o.

Applications
Bioreactors

Software-defined control for a bioreactor — a new QB vessel, an upgrade to one you have, or a retrofit of the existing PLC.

Automated sampling

Automated sampling from 4–18 sources, aseptic-capable and up to 72 hours unattended. Works with any vendor's bioreactor.

Buffer & media preparation

Automated preparation of growth media and process buffers, so a recipe runs the same way every time without fixed infrastructure.

Deployment
Retrofit

Existing equipment keeps running; QB takes over the PLC, or reads from it without touching control.

Scale
Pilot (50–300 L)

Stainless steel, where QB supplies the control software and integration and a certified partner builds the installation.

Digital maturity: today and target

Scored out of 100 across six dimensions. The target is what UPSIDE Foods's own published ambition implies — not a perfect score.

Source: A4BEE analysis of public sources
IT/OT convergence 20 → 80
Biological data from R&D and the PLC/SCADA layer governing the bioreactors sit in separate estates, so mechanical set-points are held independently of live cell conditions. Closing that loop is ahead of the company rather than behind it.
Bioprocess digital twin 18 → 85
Scale-up from bench to 2,000 liters is currently bridged through physical runs. A CFD and metabolic model of the vessel would let parameters be narrowed in simulation before media is committed to a batch.
Line automation and robotics 25 → 78
Whole-cut production runs on manually handled roller bottles and suspension culture needs tighter automated environmental control; collaborative-robot handling and an MES are the step from operator-bound output to line throughput.
Unified data platform 22 → 80
R&D, manufacturing and quality data sit in separate ELN, LIMS and control systems, so cross-cutting questions are answered by reconciling systems rather than querying one model.
Quality and regulatory systems 30 → 85
The FDA/USDA split calls for one continuous, auditable thread from biopsy to package; where genealogy is compiled per review, a continuous digital record makes real-time batch release and root-cause analysis properties of the system.
Lucius Labs B2B systems 15 → 78
The January 2026 spinoff needs customer portals, formulation software and a validated QMS to serve life-sciences clients, which is a different shape from food-production infrastructure and largely still to be built.

Check this yourself

Our Service Portal has free self-assessments and market comparisons. These are the ones that line up with what we've read above — no sales call required.

Think we've read this right?

Talk to us

Related reading

This is an independent analysis prepared by A4BEE from publicly available information as of March 2026. It reflects A4BEE's own interpretation and opinion, is not affiliated with, endorsed by, or verified with UPSIDE Foods, and may be incomplete or inaccurate. All company names and trademarks are the property of their respective owners. To request a correction or removal, contact [email protected].