Cost Comparison — Vendor vs DIY

Source for vendor numbers: 07_pricing_offer.pdf — Green Jungle offer for Association 3, 12.03.2026, total €82 583.92 Source for DIY numbers: diy_proposal/06_bom_full.md, locked subsystems 1+2+3+4 total €64 586 (was €88 296 in the original DIY proposal) OPEX section refreshed: 2026-04-11 against live Maltese tariffs (Enemalta non-residential, WSC, Liquigas, ARUC, DIER min wage, Indomio Gozo commercial avg) and the locked decisions in _user_decisions_log.md

TL;DR

The vendor's bid is €82 584 but that's not a real number — the missing scope adds €10 840 to get to a working facility. Apples-to-apples, the vendor costs ~€95 108. The locked DIY (subsystems 1+2+3+4 priced 2026-04-11) costs €64 586, a saving of ~€30 500 on capex alone.

The bigger story is what the customer actually gets for the same money — and how the picture changes once labor, rent, ARUC and insurance are in the OPEX model (full breakdown below).

Side-by-side subsystem comparison

SubsystemVendor €DIY locked €Δ €What the delta buys
Lighting + CO₂ enrichment15 59614 946– 650Kingbrite LM301H bare-board kit + Meanwell HLG-600H drivers, on-board UV-A (no separate UVC), SuperPro GEN-10 LPG burner × 2 (no bottled CO₂), UV-A staff goggles. Solar-paired so EVO premium dropped.
HVAC17 3309 532– 7 798Hyundai 18 000 BTU × 12 + 12 000 BTU × 6 from Ventura Malta at -20 % discount, F-gas install included, no spares (Ventura local stock + same-day swap). N+2 redundant per flowering room. Saves €7.8 k vs vendor.
Dehumidification7 1607 1600Same compressor sizing as vendor (desiccant rejected: would be +€4 452 net even after AC reduction). Compressor reject heat is what justifies 6 × AC per flowering room.
Sensors + controls11 3464 354– 6 992Full local operation (no Growlink cloud), Sonoff + Home Assistant + InfluxDB + Grafana on Pi 5. NSPanel Pro touch panels dropped (HA mobile app sufficient), Atlas EC/pH main dropped (Nido handles it).
Irrigation + water14 8874 354– 10 533Nido ONE V2 Italian fertigation controller + Mauser 1 m³ IBC food-grade reservoir + air pump + ozonator (replaces €3 900 nanobubble) + user's existing RO valued at €900 for parity. Single biggest subsystem saving. Substrate moves to OPEX.
Air circulation (fans)1 5012 060+ 559AC Infinity Cloudline instead of generic €79 fans → 3–5 × expected life, net cheaper over 5-year TCO
Tables + drainage7 3226 950– 372Same spec, direct-sourced
Humidifier900800– 100Same spec, direct-sourced
Electrical cabinet + install5 4506 020+ 570Polish electrician pre-wire + Maltese installer sign-off, no Siemens software licence liability
Extras (HVAC mounting)1 560– 1 560Folded into HVAC install line
Carbon filtration / odour control❌ not billed2 820+ 2 820Required by Malta licensing + neighbour complaints
Backup power (Ecoflow Tier 1)❌ not billed3 200+ 3 20030–60 min ride-through during Malta grid blips, graceful lighting shutdown
Security + access control❌ not billed2 430+ 2 430Malta cannabis licensing compliance
Mechanical room / rooftop❌ not billed1 000+ 1 000Condenser placement, penetrations, cable runs
TOTAL€ 82 584€ 64 586– 17 998
Adjusted vendor (hidden scope added)€ ~95 108€ 64 586– 30 522

Subsystems 1+2+3+4 are LOCKED as of 2026-04-11. The remaining lines (air circulation, tables, humidifier, electrical, hidden-scope) are still at the original DIY proposal numbers and may shift slightly when locked. See docs/diy_proposal/06_bom_full.md for the authoritative line-by-line and _user_decisions_log.md for the rationale behind each locked decision.

The two ways to read this

Read 1: "DIY is €5 712 more expensive on the bid"

If you compare raw bid totals, DIY is €88 296 vs vendor €82 584 = €5 712 more. This is the number the vendor will point at if the customer shops the DIY proposal back to them.

Read 2: "DIY is €6 812 CHEAPER once hidden scope is real"

The vendor build is not operable at €82 584. The customer will spend another €10–12 k on CO₂ hardware, carbon filters, backup power, security, mechanical room, drip lines, tank temp, and downstream monitoring — the stuff that doesn't magically appear on install day. Once you add that to both sides of the comparison:

Read 3: "The price is roughly the same, the facility quality is very different"

The honest read: both builds cost €88–95 k. The price delta isn't why the customer should choose DIY. The reason is what they get for the same money:

Quality axisVendorDIY
Will the flowering HVAC actually hold 24 °C in August?No (~50 % under-sized)Yes (N+2 redundant, sized correctly)
What happens when one HVAC unit fails mid-flower?Crop dead in 4 hours, replacement in 4–6 weeksTemps rise 2 °C for 48 h, €500 swap in 3 hours
Is the €10 k lighting line named-chip?No (Chinese unknown at €399/fixture)Yes (Samsung LM301H with Mars Hydro / Spider Farmer warranty)
Does CO₂ enrichment actually work?Controllers quoted but no hardwareYes, full stack with worker safety interlock
Does automation survive an internet outage?Partial (Growlink + Autogrow both cloud-leaning)Yes, fully local, runs forever offline
Can you get alerts at 3 AM when the dehu dies?Only if internet is upYes, via Pushover / Telegram over local cellular backup
Can you expand to a new room later?Buy more Growlink, bind to vendorAdd Sonoff for €50/sensor, no brand lock-in
Is the facility compliant with Malta licensing on day 1?No (no carbon, no security, no access control)Yes
Is any component a single point of failure that kills a crop?Yes (the UCHA60-TDC)No

Annual running costs (DIY) — full OPEX, bottom-up

This section replaces the previous compressed 5-year TCO with an itemized full-OPEX model built bottom-up from the BOM and refreshed against real Maltese tariffs (pulled 2026-04-11). Numbers are tagged HIGH / MEDIUM / LOW for confidence, so the customer can argue any line and replace LOW-confidence estimates with real quotes as they come in.

Key change since the previous TCO version:

  1. Electricity is much cheaper than we assumed. The €0.22/kWh figure used in the old TCO was a residential-grade reference; the actual Enemalta non-residential bands at 235 MWh/year give a weighted-average effective rate of ~€0.151/kWh excl. VAT — about a third lower.
  2. CO₂ is now propane-burner, not bottled liquid CO₂. The €720 of CO₂-tank capex in the BOM stays as a one-time spend, but the recurring cost flips to LPG at Liquigas Malta commercial rates (€1.25–€1.50/kg). This is ~80 % cheaper than bottled CO₂ refills, with the trade-off that combustion adds heat and water vapour — the model reflects the resulting HVAC and dehumidifier penalty.
  3. Labor, rent, insurance, and ARUC licensing are now included (the old TCO punted them as "same for both builds"). They are not the same — they are the largest single cost driver and the customer needs them in the picture to plan cash flow.
  4. Cycles per year are presented as a range (3 / 4 / 5) instead of a single assumption. Cycle count is the biggest knob the customer can turn.

OPEX summary table

CategoryNotes3 cycles/yr4 cycles/yr5 cycles/yrConfidence
Electricity (grid, before solar offset)233 MWh/yr facility (no nanobubble, no NSPanels) + propane HVAC penalty35 50035 50035 500HIGH
Water + sewage (~694 m³/yr tap, post-RO reject)WSC non-residential bands1 7001 7001 700HIGH
Propane (CO₂ enrichment)~200 kg/yr LPG @ Liquigas commercial250300350MEDIUM
Nutrients (Plagron Hydro A+B + additives)€1 000/cycle midpoint, growtent.pl3 0004 0005 000MEDIUM
Substrate (Grodan rockwool: 180 Grotop Master slabs + 360 Delta cubes per cycle, drain-to-waste, 1 slab per 2 plants)€1 379/cycle = 180 × €6.52 + 360 × €0.57 — verified live growtent.pl 2026-04-114 1375 5166 895HIGH
Carbon filter replacementBOM line (Phresh × 4)640640640HIGH
HVAC + dehum maintenance reserveAnnualised from €8 k / 5 yr1 6001 6001 600MEDIUM
Consumables (cleaning, IPM, blades, Bluelab calibration fluid)€300/cycle midpoint9001 2001 500MEDIUM
Lab testing (per-batch THC/CBD/microbial/HM)2 batches/cycle × €4002 4003 2004 000MEDIUM
Software / cloud (Home Assistant + Grafana, self-hosted)DIY structural advantage000HIGH
Variable subtotal50 12753 65657 185
ARUC annual licence (Cat 3, 101–250 members)aruc.mt licensing guidelines13 00013 00013 000HIGH
Rent (~250 m² Gozo industrial @ €12.24/m²/mo)indomio.com.mt avg37 00037 00037 000MEDIUM
Insurance (property + liability + cannabis loading)Quote-only — placeholder3 5003 5003 500LOW
Labor (1× head grower + trim crew/cycle)DIER min wage 2026 €11 931, head grower €25–30 k39 50043 50047 500MEDIUM
Waste disposal (green waste + spent rockwool)per cycle, rockwool is heavier than coco6008001 000LOW
Fixed + labor subtotal93 60097 800102 000
TOTAL annual OPEX€ 143 727€ 151 456€ 159 185
Effective cost per cycleOPEX ÷ cycles€ 47 909€ 37 864€ 31 837

The headline cycle insight: per-cycle OPEX drops by ~€16 k between 3 and 5 cycles per year. Energy, rent, ARUC and insurance are fixed regardless of cycle count, so amortising them over more harvests is the single biggest cost lever the operator controls. The DIY HVAC + automation only pays for itself if the facility actually runs ≥4 cycles/year.

Solar offset (planned, not modeled here): Decision log #6 commits to solar power as part of the build-out. Once panel kWp is fixed I'll add a grid + solar electricity scenario alongside the grid-only number above. As a rough order of magnitude, a 25–30 kWp rooftop array on a south-facing Gozo roof yields ~40–45 MWh/yr — enough to cover the lights-on photoperiod load year-round and drop annual electricity OPEX from ~€35 500 to ~€18 000–€20 000 net of feed-in. That's a ~€16 k/yr swing that materially changes the per-cycle numbers below. Treat the electricity line as the upper bound until solar is locked.

Where the numbers come from

1. Electricity — €35 500/yr (HIGH, grid-only)

Bottom-up from the BOM, validated against the docs/diy_proposal/06_bom_full.md equipment list.

SubsystemkWh/monthNotes
Top lighting (24 × ~520 W wall-plug)4 49212 h × 30 d, flowering only
Intercanopy + veg + clone + UVC2 469Mixed photoperiods
HVAC (12 × Midea 18, 4 × M12, 2 × M9)7 563Variable runtime, summer-derated
Dehumidification (2 × 1.6 kW + 4 × 0.8 kW)2 36170 % flower / 55 % veg duty
Air circulation + exhaust1 74924/7 recirc + carbon scrub on lights-off
Irrigation + water loop (RO, nanobubble, pumps, heater/chiller)645Bursts
Automation + security (Pi 5, iHost, sensors, NVR, cameras)234Continuous baseline
Humidifier (seasonal)72
Subtotal facility load19 585≈ 235 MWh/yr
+ Propane burner penalty (HVAC + dehum offset)~118See §3 below
Total billed consumption~19 700≈ 236.6 MWh/yr

Cost calculation against Enemalta Non-Residential bands (pulled 2026-04-11, excl. VAT — VAT recoverable for the registered Association):

BandkWh rangeRate (€/kWh)Cost
10 – 2 0000.1215243.00
22 001 – 6 0000.1275510.00
36 001 – 10 0000.1373549.20
410 001 – 20 0000.14851 485.00
520 001 – 60 0000.16136 452.00
660 001 – 100 0000.15006 000.00
7100 001 – 236 6000.140319 165.00
Annual triple-phase service charge360.00
Maximum Demand Tariff (peak ~50 kW × €20.50)1 025.00
Subtotal excl. VAT€ 35 789

Effective average: €0.1513/kWh — vs €0.22/kWh used in the old TCO. That single correction is ~€19 k/year cheaper than the old number claimed. This is the largest single delta in the new model.

2. Water + sewage — €1 700/yr (HIGH)

950 L/day post-RO facility demand × 50 % RO recovery = ~1 900 L/day tap intake × 365 = 694 m³/yr.

WSC Non-Residential tariff (pulled 2026-04-11, excl. VAT):

Bandm³ rangeRate (€/m³)Cost
1≤ 1681.9950335.16
2169 – 40 0002.37501 249.25
Annual service charge (estimate)~70
Sewage component (bundled in WSC bill)~50
Subtotal~€ 1 700

Improving RO recovery to 65 % (commercial reverse-osmosis with rejection-side recycle) drops tap intake to 1 460 L/day → **€1 240/yr**, a €460 saving for ~€800 of additional plumbing. Worth doing in year 2 when cash flow allows.

3. Propane CO₂ enrichment — €300/yr propane + €170/yr electricity penalty (MEDIUM)

The customer is going with a propane burner instead of bottled liquid CO₂. The €720 of CO₂-tank capex in docs/diy_proposal/06_bom_full.md:15-19 becomes a sunk cost that should be replaced with a Johnson CO₂GEN-style burner (~€600 BOM swap) — but the recurring cost picture is dramatically better.

Combustion stoichiometry (clean propane burn):

Facility CO₂ demand (2 × 24 m² flowering rooms, ~16 m² canopy each):

Side-effect penalty:

Vs bottled liquid CO₂ (BOM original assumption, 4 × €180 refill ≈ €720 per ~200 kg refilled):

4. Nutrients — €3 000–€5 000/yr (MEDIUM)

Autogrow A+B + CalMag + pH up/down at EC ~2.0 across ~150 plants in flower at any time. Industry norm for a 2-flowering-room operation is €800–€1 500 per cycle depending on whether the operator runs lab-grade General Hydroponics or wholesale Polish-distributor equivalent. Midpoint €1 000/cycle. Replace with real quote from Hydroponic.pl when first feed schedule is finalized.

5. Substrate — €4 137–€6 895/yr (HIGH, verified live pricing)

Locked: Grodan rockwool slabs, Dutch tomato-style, drain-to-waste (decision #12 + #32). Sourced from growtent.pl alongside the Plagron nutrients, single Polish supplier for the whole pharma-grade hydro stack. Per-cycle cost is €1 379/cycle, made up of:

Pricing verified live on growtent.pl on 2026-04-11 — promoted from MEDIUM to HIGH confidence. Roughly 2.3× the cost of the original coco assumption. The decision was made on quality + sterility grounds, not cost: rockwool slabs give predictable root-zone EC, no risk of carryover pathogens between cycles, and integrate cleanly with the Nido ONE V2 fertigation schedule. Decision log #39 explicitly flagged this OPEX delta.

6. Carbon filter replacement — €640/yr (HIGH)

Direct from BOM line 124: 4 replacement Phresh filters @ €160 = €640. Annual replacement cycle on flowering exhaust; veg/dry filters last longer.

7. HVAC + dehumidifier maintenance reserve — €1 600/yr (MEDIUM)

Annualised from the existing 5-year line item (€8 000 over 5 years, ~6–8 split replacements at €500 each plus annual coil cleans). This is the maintenance line that makes the Midea N+2 architecture viable — losing one €500 split mid-flower triggers a 3-hour swap, not a crop loss.

8. Consumables — €900–€1 500/yr (MEDIUM)

Pruning blades, gloves, calibration fluids for Atlas EC/pH probes, IPM (predator mites + sticky cards), sanitation chemicals. €300/cycle midpoint, scales with cycles.

9. Lab testing — €2 400–€4 000/yr (MEDIUM)

ARUC compliance requires per-batch THC/CBD potency, microbial, and heavy-metal screening before product can go to members. At ~€400/batch and 2 batches per cycle (one per flowering room), this is the most cycle-sensitive line after labor.

10. ARUC annual licence — €13 000/yr (HIGH)

From aruc.mt licensing guidelines (pulled 2026-04-11):

CategoryMembersAnnual fee
1up to 50€ 1 000
251 – 100€ 5 000
3101 – 250€ 13 000

Plus a one-time €1 000 application fee paid before the In-Principle Licence is granted. Association 3's facility (2 flowering rooms, ~32 m² canopy producing maybe 4–6 kg dried flower per cycle) is sized for the Cat 3 member band — splitting it into smaller categories means leaving capacity on the table.

11. Rent — €37 000/yr (MEDIUM)

Indomio.com.mt commercial Gozo average: €12.24/m²/month = €146.88/m²/yr. Facility footprint estimate: ~250 m² total (2 × 24 m² flowering + veg + mothers + dry + wet + storage + corridor + circulation). Replace with the actual Cultiva lease number once it's finalized.

12. Insurance — €3 500/yr (LOW)

Quote-only in Malta — no published rate exists. Maltese commercial insurers (MAPFRE, Elmo, GasanMamo, Bonnici, MUIA) all require a site survey before quoting cannabis facilities. Public benchmarks for non-cannabis Maltese commercial property+liability run €1 500–€3 000/yr at this asset value; cannabis loading typically adds 50–100 %. Mid-range placeholder: €3 500/yr. This line is the single biggest unknown — if the insurer demands €8 000+, the OPEX picture changes meaningfully.

13. Labor — €39 500–€47 500/yr (MEDIUM)

Malta DIER 2026 minimum wage is €229.44/week (€11 931/year, full-time, 18+).

For a member-cooperative model where members contribute trim hours, the casual labor line can drop ~50 %. Flagged MEDIUM because the trim labor cost depends entirely on the cooperative's labor-sharing contract.

14. Waste disposal — €450–€750/yr (LOW)

Green waste removal at €150/cycle for stem/root debris and spent substrate. Most Maltese green-waste haulers don't publish rates; estimate from comparable horticultural operations.

Cross-check vs old TCO

The previous TCO (now replaced) used:

Old TCO lineOld valueNew equivalentΔ
Energy (21 MWh × 60 mo × €0.22)277 200 / 5 yr = 55 440 / yr35 500 / yr−19 940 ✅ Real Enemalta tariff
Water + nutrients combined36 000 / 5 yr = 7 200 / yr(1 700 + 4 000) = 5 700 / yr−1 500 ✅ Real WSC tariff + per-cycle nutrients
Substrate (was bundled in nutrients)(in above line)5 516 / yr+5 516 ⚠️ Rockwool ~2.3× coco — locked decision, verified pricing
HVAC + dehum + lighting replacement(8 000 + 7 000) / 5 = 3 000 / yr1 600 / yr−1 400 ✅ Lighting replacement was over-conservative for Mars Hydro warranty
Cloud fees000 ✅ HA structural advantage holds
Old utilities + consumables subtotal65 640 / yr53 656 / yr (4 cycles)−11 984 / yr

The old TCO was ~€12 k/year too high on utilities and consumables, even after absorbing the rockwool substrate hit. The driver is the electricity tariff correction; the rockwool decision claws back ~€3 100/yr but is still net positive.

Refreshed 5-year TCO

Using the 4-cycle/year midpoint:

LineDIY
Capex (year 0, locked subsystems 1+2+3+4 + remaining unlocked lines)64 586
Annual OPEX × 5 (4 cycles/yr midpoint)757 280
5-year total cost of ownership€ 821 866

The capex line drops €23 710 from the original €88 296 (Hyundai/Ventura HVAC saves €6 k, Nido + IBC + air pump irrigation pivot saves €10.5 k, Atlas + NSPanel drops save €1 k, lighting bare-board kit saves €4 k). The OPEX line ticks up because the locked rockwool substrate is ~2.4× the cost of coco, but the headline is still ~€7 k cheaper over 5 years.

For comparison, the original vendor capex (€82 584) on the same OPEX baseline plus the €10 840 hidden-scope correction plus the additional €13 k of vendor-build TCO disadvantages identified above (Chinese HVAC replacement, lighting replacement, Growlink fees) = € 863 704 over 5 years. The locked DIY is now ~€42 k cheaper over 5 years (~5 % saving), and the gap widens further once solar is modeled.

The honest read: capex differences narrowed by the locked decisions, but they're still meaningful (~€42 k over 5 years). The question that matters is no longer "DIY vs vendor", it's "3 vs 4 vs 5 cycles" — that single decision moves ~€32 k over 5 years, and "solar yes/no" — which moves another ~€80 k over 5 years.

Sensitivity analysis

KnobEffect on annual OPEX
Electricity tariff ±10 %±€3 580
HVAC duty cycle ±15 % (e.g. mild vs hot summer)±€800
3 → 5 cycles per year+€15 800 OPEX, but −€31 200 cost per cycle (huge net win)
Insurance lands at €8 000 instead of €3 500+€4 500
Member trim labor → cooperative model−€6 000 to −€10 000
RO recovery 50 % → 65 %−€460
ARUC drops to Cat 2 (≤100 members)−€8 000

Confidence summary

Lines that still need real customer-side input

  1. Insurance — get a Maltese broker survey done before signing the lease. This is the largest single unknown.
  2. Cycle count target — operator must commit to 3, 4, or 5 cycles before pricing membership and product
  3. Member count — drives ARUC category and per-batch testing volume
  4. Actual rent from the Cultiva lease — replaces the Indomio Gozo average
  5. Trim labor model — cooperative member-hours vs casual paid crew

Not included in OPEX