China still produces roughly 28% of global manufacturing output, and the Pearl River Delta cluster (Shenzhen, Dongguan, Zhongshan) holds the densest LED packaging and driver-IC supply chain on the planet, which is why direct LED sourcing from Chinese factories typically lands 30 to 50% below domestic procurement once the full landed cost is calculated [S2].
That price gap only survives if you run a structured qualification loop: lock the photometric and driver spec, separate the plant from the trading company, demand LM-80 and IES test data, and inspect a pre-shipment sample against your own integrating-sphere reading, not the supplier's PDF [S1][S2].
Where the LED supply chain actually sits, and why the cluster matters
LED production in China is geographically concentrated: Shenzhen and Dongguan dominate LED packaging, driver electronics, and SMT assembly, while Zhongshan (Guangdong) still holds a meaningful share of finished residential and commercial luminaires; this co-location of chip-on-board lines, phosphor suppliers, and driver-IC vendors inside a roughly 150 km radius is what keeps 7 to 14 day rapid-turn lead times workable for small OEM runs [S1][S5].
For buyers, that cluster density matters more than headline FOB price, because it shortens the iteration loop on samples and lets you run a re-spin on a driver or beam angle without restarting a 45 to 60 day production calendar [S1]. When a quote arrives unusually cheap, it usually signals a trader, not a plant, sourcing sub-binned LEDs from a non-cluster secondary supplier [S1].
Spec lock: the eight items you write down before the first RFQ
Maple Sourcing's category-failure analysis is blunt about it: suppliers working from different definitions of your product are not quoting the same job, and that is exactly where LED projects derail, since one bidder assumes a 90 CRI COB, another a 80 CRI SMD array, and a third a different dimming protocol [S3].
Lock these eight items before the first inquiry, and you eliminate roughly 70% of the sample-versus-production drift that drives returns: (1) target market and its mains voltage (120 V North America, 230 V EU, 220 V most of Asia), (2) CCT and tolerance bin (e.g. 3000 K ±150 K, or 3-step MacAdam), (3) CRI minimum (Ra ≥ 80 for hospitality, ≥ 90 for retail/museum), (4) luminous efficacy floor in lm/W at the driver, not the LED, (5) dimming protocol (0 to 10 V, DALI-2, Casambi, TRIAC), (6) IP and IK rating for the luminaire housing, (7) photometric file format (IES, LDT, or both), and (8) driver certification list (UL, CE, ENEC, CCC, KC, PSE) for the destination market [S3][S4].
Supplier-type decision: factory, trading company, or sourcing agent
Three options exist on every LED RFQ, and each fits a different buyer profile, so match the channel to your order profile, not to the cheapest quote: [S3]
Direct factory works when you can spend 7 to 14 days on a pre-shipment inspection and you have a finished spec; a plant running OEM/ODM gives the lowest unit cost and the only path to a custom driver PCB, but expects 1,000 to 5,000 unit MOQs and a 30 to 45 day production lead [S1][S5]. Trading company is the right channel for first-time buyers placing sub-1,000-unit mixed-SKU orders under, say, USD 10,000, where the trader absorbs the MOQ fragmentation and re-sells from stock at 8 to 15% above factory FOB [S1][S5]. Sourcing agent (typically 5 to 8% of order value) is justified once you are past your second or third reorder and want independent QC, bilingual negotiation, and a buffer against the well-documented sample-versus-mass-production drift that hits LED bins hardest [S1][S2][S5].
Vetting: four documents and two tests that actually filter
Four documents separate a legitimate LED supplier from a catalog reseller: (1) a verifiable business license with a real factory address, not a co-working space; (2) an LM-80 report for the LED package (not just the module), typically a 9,000-hour TM-21 projection; (3) a driver certificate matching the destination market (UL Class 2, ENEC, CCC, or PSE); and (4) a recent IES file traceable to a real goniophotometer run, not a simulated export [S3][S4][S5].
Two tests catch what documents hide. First, request a 5 to 10 unit pre-production sample and measure CCT, CRI, and luminous flux on your own integrating sphere or a calibrated handheld spectrometer; suppliers' internal datasheets routinely overstate efficacy by 8 to 15% [S2][S3]. Second, run a 48 to 72 hour burn-in at the rated driver current and ambient, and check for early catastrophic failures or color shift above Δu'v' 0.005, which is the threshold where most hospitality clients reject a lot [S3].
Cost, MOQ, and lead-time bands to anchor your RFQ
For OEM LED luminaires in 2026, realistic bands based on the cluster pricing data: a private-label 600 mm LED batten at 20 to 30 W sits in the USD 3.50 to 6.00 FOB range at 1,000-unit MOQ, dropping to USD 2.80 to 4.50 at 5,000 units; a 100 to 150 W UFO highbay at 1,000 units lands around USD 18 to 28 FOB, and a 9 to 12 W residential downlight at 1,000 units runs USD 2.20 to 3.80 FOB, with driver and 0 to 10 V dimming included [S2][S4].
MOQ is the second negotiation lever after price: most Chinese LED plants quote 500 to 1,000 units per SKU for finished luminaires but will drop to 200 to 300 units if you accept a standard driver, neutral white 4000 K only, and a stock housing color [S2][S5]. Production lead time runs 25 to 35 days after sample approval for 1,000 to 3,000 unit runs, plus 5 to 7 days for export documentation, and 18 to 32 days ocean freight to North America or 28 to 40 days to the EU under FOB Shenzhen [S4].
Shipping, Incoterms, and the 2026 freight reality
For most buyers, ocean FCL or LCL out of Shenzhen or Yantian is the default and the cheapest per CBM, but it ties up working capital; air freight makes sense for reorders under 500 kg or for sample-to-production gap-bridging, typically at USD 4.50 to 8.00 per kg for general cargo to North America or EU [S4].
The Incoterm choice changes your risk surface more than people realize: EXW puts the entire logistics chain on the buyer, which sounds cheap but exposes you to inland trucking delays and Chinese export documentation errors; FOB is the practical default for buyers with a forwarder; DDP is the right call only if you sell on platforms like Amazon FBA or Shopify and have no US import setup, accepting the 12 to 18% all-in premium [S4]. DDP-linked 2026 guidance for adjacent electronics categories, including the EV charger sourcing pathway covered in this spec map, runs the same Incoterm economics and is a useful cross-reference.
Compliance by destination: where projects actually get seized
Non-compliance, not price, is the single largest cause of LED shipment failures at destination. North America requires UL or ETL on the driver, FCC Part 15 on the LED module, and Energy Star or DLC listing if you want to sell into utility rebate programs; the EU requires CE under the Low Voltage Directive 2014/35/EU and EMC 2014/30/EU, plus RoHS 2 (2011/65/EU) and (EU) 2019/2020 ecodesign for light sources; the UK requires UKCA after the transition period; and most of the Middle East, India, and Southeast Asia accept CE but layer on local marks like KC for South Korea and PSE for Japan [S3][S4].
Build a compliance file before the first deposit, not after the first shipment: certificate PDFs, the LM-80 report, the IES file, the DoC (Declaration of Conformity), the RoHS test report, and a sample of the driver label, all dated within 24 months, all from accredited labs with accreditation numbers visible [S3].
Quality control gates that actually catch LED defects
Three inspection gates separate a clean project from a claim dispute, in this order: (1) during-production inspection (DUPRO) at 20 to 30% completion, checking driver soldering, LED bin consistency, and housing IP gasket integrity; (2) pre-shipment inspection (PSI) on 100% of finished goods, with AQL 2.5 sampling for cosmetic defects and AQL 1.0 for electrical/photometric; and (3) a 1 to 2% container loading supervision to catch carton count, label mix-ups, and the classic "wrong color temperature in the wrong carton" defect that costs more in returns than the inspection itself [S1][S2][S3].
For buyers without an in-house QC team, a third-party inspection runs USD 200 to 350 per man-day in Shenzhen, and pays for itself on the first lot where it catches a driver batch with the wrong constant-current IC, which is the single most common LED failure mode seen in 2025 to 2026 claims data [S1][S2].
Comparison: factory vs trading company vs sourcing agent on five buyer criteria
Trackable next signals for the LED sourcing cycle
Two signals to watch into Q4 2026: the Made-in-China.com buyer-activity index, which printed +25% in April 2026 and is the single best free demand thermometer for cross-border LED sourcing [S2], and the UL/ENEC driver-certificate queue, which currently sits at 6 to 8 weeks for new submissions, so if your Q1 2027 launch needs the cert, the spec freeze has to land before October 2026. For an adjacent view on cluster-based electronics sourcing economics, the industrial display price-to-spec map covers a similar Pearl River Delta cost curve and is worth reading alongside this guide.
Spec-level background on the components involved: linear guide, crossed roller guide, and pressure transmitter.