Commodities trade on international benchmarks. The collateral already sits, graded and insured, in a warehouse. We verify and tokenize that collateral and bring it to lending protocols that already exist — we don't build a lender. The farmer refinances 18–25% BRL debt at ≈11% in dollars — while the export leg funds at ≈8% in RMB. This deck maps the build — and the partner seats it needs.
18–25% → ≈11%
farmer's BRL rate vs USD all-in, refinanced at the warehouse
8%
the investor's target APR, in dollars — senior & insured
50–60% LTV
launch range · 65% the ceiling with full checks — 80% is the buy-back level, not an LTV
~0.07%
of farm-credit notes tokenized — the rail is unbuilt
Alexey Karanyuk · Giants Labs · Florianópolis · not an offer of securities · not investment or legal advice
02
Contents — from the problem to the ask
The logic: the scheme in one picture (p. 3) → the problem (the world's farm borrows at 18–25%) → what we do (tokenize the stored grain, refinance through global lending protocols) → why it doesn't work yet (the warehouse black box) → the build (three pillars: operational · tokenization · financial) → market & competition → stakeholders & the ask.
01
The problem
The world's farm at 18–25% · where the exports go · one loan wraps the whole cycle
p. 4–6
02
What we do
The collateral rail · one cycle, two loans · cycle risk vs shelf risk
p. 7–9
03
Why it doesn't work yet
The warehouse black box · the unmeasured receipt rail
p. 10–11
04
The build — three pillars
Pillars · architecture · interface · two layers · capital rail · oracle · default day · liquidator · FX · economics · the RMB leg
p. 12–22
05
Market & competition
Tokenization already works — all of it whole-cycle · how our offer differs
p. 23–24
06
Stakeholders, questions, partners & contact
Four seats: want ↔ need · open questions · the ten-seat map · the ask
The scheme — why credit is expensive, and what makes it cheap
Six links, one picture. Credit is expensive because the collateral behind it is illiquid; the one liquid asset — stored grain — doesn't work as collateral yet; we make it work, bring the lenders, and the rate falls. The rest of the deck proves each link.
todaythe problem chain
1 · Credit is expensive
The world's farm borrows at 18–25% in BRL — one whole-cycle loan, every risk of the season priced into a single rate (p. 4–6).
→
2 · Because collateral is illiquid
Behind the rate sits collateral a lender can't sell fast — land, machinery, next season's crop. Slow to seize, slower to sell — so every loan is priced as if barely secured.
→
3 · The liquid asset sits idle
Stored grain is graded, insured and dollar-priced — the best collateral in the country. But the warehouse is a black box, so it barely works as a pledge (p. 10–11).
↓ WHAT WE CHANGE
4 · We make the grain work
Digitize — continuous, independent proof the grain is there · tokenize — the electronic warehouse receipt (CDA/WA) becomes a clean, fundable title · a liquidator network — standby buyers turn a seized lot into cash in 24–72 h (p. 12–19).
+
5 · And bring the lenders
Global USD pools plug into collateral they can finally verify — plus an RMB leg at the export side, where Brazil's #1 soy buyer lives (p. 16, 20–22).
=
6 · Cheap credit
The storage leg reprices to ≈11% USD all-in from 18–25% BRL — the pre-harvest leg stays local (p. 8–9); the investor earns 8% senior, insured (p. 21). Targets, not promises.
Giants Labs · Brazil warehouse-credit · the causal chain — each link is a slide · 03 / 28
04
Brazil: the world's farm — and it borrows at 18–25%
The scale first: Brazil is the #1 exporter of soy, beef, coffee, sugar and cotton — $169 bn of agribusiness exports in 2025, all of it dollar-priced. The pain second: the machine that grows it borrows in reais at 18–25%.
🇧🇷 Export revenue by product — 2025 (US$ bn)
cotton revenue = 2024/25 commercial year (Abrapa); other figures calendar 2025
$169 bn
agribusiness exports in 2025 — 48.5% of everything Brazil sells abroad.
18–25%
the BRL rate the farm machine actually borrows at — the gap this deck attacks.
And it all passes through warehouses
A 358.6 Mt harvest vs 233.8 Mt of storage — Brazil can store barely two-thirds of what it grows, and ~85% of the crop moves through third-party warehouses (on-farm just ~15% vs 65% in the US). The paper that flow generates — warehouse receipts — is the collateral base this deck is about. Where the revenue goes — next slide.
Where it goes — every region buys; one buyer takes a third
The $169 bn doesn't ride on one market — every region buys Brazilian agro. But the concentration is real: China (+HK) alone takes 33% of the total and ~80% of soybean exports — exactly why the funding design runs a USD base with an RMB leg at the export side (p. 21–22).
Concentration in one number
~80% of 2025 soybean exports went to China (ANEC) — up from a 74% average in 2021–24.
How it's financed today — one loan wraps the whole cycle
Farm credit runs through whole-cycle instruments: the CPR — a pre-harvest promise covering planting to sale — with securitized wrappers (CRA/CDCA) on top. CPR alone is R$560 bn, ~3.3× since 2022, R$565 bn by May 2026. One loan, one 18–25% BRL rate — every risk of the cycle bundled inside it.
Outstanding stock by instrument — Mar 2026 (R$ bn)
CPR registered stock — trajectory 2022 → 2026 (R$ bn · x-axis not to time scale)
Bundled into that one rate: drought, crop failure, execution, price, counterparty — the whole cycle priced as one risk. Unbundling it is the opportunity this deck is about.
Giants Labs · Brazil warehouse-credit · sources: MAPA Boletim de Finanças Privadas do Agro · B3 · ANBIMA · 06 / 28
07
What we do — tokenize the stored grain, plug it into lending protocols
We take the warehouse receipt on grain already in store, verify it, tokenize it, and deliver it to lending protocols that already exist. The farmer refinances 18–25% BRL debt at ≈11% in dollars. We never hold the loan — one layer changes: where the money comes from.
Six layers of the deal — five we keep, one we swap
The title, its registration, its risk wrap and its enforcement stay exactly as the Brazilian courts know them. We swap the funding layer only — a local bank balance sheet becomes a global USDC pool — and we run the data that makes it fundable: monitoring, audits, lien checks, the price feed. No credit desk, no balance sheet, no borrower underwriting — the lender of record is always someone else (the same line BCB Res. 520 art. 12 draws for VASPs).
Giants Labs · Brazil warehouse-credit · the collateral rail — we are never the creditor · 07 / 28
08
One cycle, two loans — where we cut in
The old world finances the green bar — one CPR loan across the whole cycle. We split the cycle at the warehouse gate: the amber CDA/WA window, when the grain is physically in store, is where our USD refinancing lives. Same cycle, two loans, two risk regimes.
Illustrative — to be verified. Per-crop day-windows have no published source; shown to convey where the amber refinancing window sits inside the full cycle, not as exact figures. At the gate, the receipt-backed USD loan pays off the pre-harvest BRL debt — the farmer stops fire-selling into the harvest-month glut.
Two risk regimes — the whole cycle vs grain on the shelf
Why we insist on the split: a whole-cycle loan prices every risk into one rate. Grain on the shelf carries almost none of them — the asset already exists, graded and insured, and it's dollar-priced by construction. Different risks deserve different capital.
LOAN 1 · The whole cycle — not our market
Drought & weather — the crop may simply not arrive
Farmer counterparty — performance over 12–18 months
Price over a long window — planting-to-sale exposure
All bundled into one 18–25% BRL rate — served by local banks, traders and barter, who are built to carry production risk.
LOAN 2 · Grain on the shelf — our market
The asset exists — graded, weighed, insured in a warehouse
Dollar-priced by construction — CEPEA/ESALQ & B3 track CBOT/ICE → no FX hedge needed · the route basis stays BRL-linked, deducted up front (p. 17)
Short window — ~180 days, storage to sale
Over-collateralised — LTV 50–60% at launch (local convention lends 60–75%)
A senior USD claim on a commodity that already exists — the safest slice of the cycle, and the one local capital overprices most.
Today both halves are priced as one. Tokenization so far copies that mistake — it wraps whole-cycle paper (p. 23). We tokenize only the shelf.
Giants Labs · Brazil warehouse-credit · production risk stays local · shelf risk goes global · 09 / 28
10
Why it doesn't work yet — the warehouse is a black box
Between the receipt and the grain sits the operational gap: the paper says one thing, the silo may hold another. Every warehouse loss on record is one of five failure modes — each with a named counter. The counters are what we install, with grain-monitoring partners (GrainTwin-type digital twins, CMAs, auditors).
Failure mode
What happens in practice
The counter we install
Phantom inventory
receipts issued, grain gone — or never there. Grupo Safras, MT 2026: R$1.78 bn, ~900 creditors, depositors' grain vanished
real-time silo monitoring (event-driven digital twin) · scheduled independent audits · collateral manager on site
Double pledge
the same lot financed twice — CPR re-issued over pledged grain ("repique", documented market practice)
registrar lien check (B3/CERC) at issuance · on-chain lien registry — the gap Drex's gravames pivot targets
Quality & commingling
bulk storage mixes lots; humidity and pests eat the grade — the black-box norm banks price in
the warehouse itself fails; enforcement freezes in court — AgroGalaxy 2024 (unsecured −85%) · essencialidade 180+ d
CDA = depositor's title, not the operator's estate · diversification · insurer subrogation · standby buyer (p. 19)
Paper–asset drift
the receipt lives on after the grain moves — the elevator-fraud archetype (operator sells everyone's grain)
token validity tied to monitored custody — the receipt dies the moment the grain leaves the walls
The scam museum runs deep — Avestruz Master's CPR-on-ostriches Ponzi (>R$1 bn, 2005), Paranaguá scale-tampering skimming, Safras's vanished soy. Every one is a control failure at the physical layer — this table exists to close them. Case files available on request.
Giants Labs · Brazil warehouse-credit · five failure modes, five counters · verification before capital · 10 / 28
11
Brazil's warehouse base is real — the receipt rail is unmeasured
Brazil's storage network is large and documented. But the layer this thesis rides — warehouse receipts (CDA/WA) — is so thin that no regulator even publishes its volume. That's not a data gap; that's the unbuilt rail.
The funnel — from storage units to actual CDA/WA issuance
Read: the infrastructure exists, the legal instrument exists — the measured, verified, investable receipt market does not. First mover builds the measuring stick.
~125 Mt
storage deficit vs the 358.6 Mt 2025/26 harvest — grain must flow through third-party warehouses every season.
Lei 15.429 · 05 Jun 2026
Warehouse certification (SNCUA) just became voluntary. The state quality filter left the field — independent collateral verification has to replace it. That's exactly the layer we build.
Nobody counts CDA/WA
MAPA's monthly agro-finance bulletin tracks CPR, LCA, CDCA, CRA and Fiagro — CDA/WA isn't even in it. BCB collects registrar data (IN 520/2024) but publishes no aggregate.
Giants Labs · Brazil warehouse-credit · sources: IBGE 2S2025 · MAPA / CNN Brasil 06.2026 · Lei 15.429/2026 · CONAB · 11 / 28
12
The build — three pillars
Everything that follows groups into three pillars: the operational layer that proves the grain, the tokenization layer that turns title into a fundable token — its legal rails included — and the financial layer that moves the money. Each pillar has its own partner seats on the map (p. 27).
01 · Operational
Prove the grain is there
Digital twin — continuous stock confirmation through channels independent of the warehouse
State machine — any encumbrance takes three confirmations: owner · warehouse · lender (p. 14)
The gate is the warehouse, not the farmer — designated warehouses admitted once; no farmer scoring
Independent audits + collateral manager on site · registrar lien checks — no double pledge
02 · Tokenization
Turn the title into a fundable token
The token carries grade × basis × delivery window — never a bare “1 tonne of grain”
Two layers — receipt NFT below, pool shares above (p. 15)
The price oracle — public benchmarks, formula basis, fail-closed; the platform never sets a price (p. 17)
Legal rails — already law
CDA/WA since 2004 (Lei 11.076) · electronic + central depository since 2022 (Lei 14.421) · bankruptcy-remote (art. 12) · extrajudicial exchange auction on default (art. 17 §2º) — no new law needed
03 · Financial
Make the money flow
The capital rail — each instrument to its own DeFi market (p. 16)
Two currencies — USD base case + an RMB leg at the export side (p. 20, 22); money legs run as câmbio through an authorized bank
The liquidator network — a standby bid, not an auction hope (p. 19)
Honest risk pricing — insurance covers physical loss, not price · LTV 50–60% · economics p. 21
The architecture — five layers, from silo to capital
One stack: physical events at the bottom, money at the top. Every token state change is bound to a physical or registry event — nothing moves on anyone's say-so. Data climbs the stack; capital and legal force come down the same five layers on default.
Events, not statements. A receipt token is minted on physical intake and dies the moment the grain leaves the walls — issuance can't outrun the silo.
No single point of trust. The warehouse reports, an independent channel measures, the registrar holds the lien, the oracle prices — no one of them can move value alone.
One data model, four views. Owner, warehouse, lender and liquidator read the same record — the interface argument, next slide.
Giants Labs · Brazil warehouse-credit · five layers · events bind the token to the grain · 13 / 28
14
The interface — the credit layer grain software is missing
Grain software is built around inventory dashboards — credit mechanics are missing: in a leading 25-screen industry package we reviewed, not one screen covered loan issuance, revaluation, margin call, default or liquidation. Our interface concept adds exactly that layer.
The liquidator console is the screen grain-inventory packages typically lack — and without it the 24–72 h window stays on paper. When the owner sees “430 t pledged”, the lender sees “coverage 1.08” — the same record, so a dispute can be about facts, never about whose system is right.
Three confirmations
No unilateral state change: owner + warehouse + lender sign every encumbrance. The one exception — liquidation: lender-initiated, on system-logged grounds only.
A third status
Between “verified” and “failed” sits “requires attention” — existing loans run on, new borrowing pauses, the warehouse gets a deadline to explain. Degradation is visible before collateral goes bad.
A score committees can audit
Not “trust index 91/100” but “7 of 10 known receipt-failure modes closed by product mechanics” — a finer cut of the five families on p. 10; two partial, one (paper issuance outside the registry) stated openly. Concept-stage count; every line is checkable.
All four role consoles exist as real, clickable markup — a build spec for the pilot, available on request.
Giants Labs · Brazil warehouse-credit · interface concept · finance-first, not inventory-first · 14 / 28
15
Two layers: the warehouse receipt below, liquidity above
Don't fractionalize the title — that breaks seizure. Fractionalize the claim on a pool of warehouse-backed loans. Capital flows up; legal force flows down to a real lot in store.
Liquidity
Small ticket, instant in/out — lives in the fungible upper token.
Seizure
Whole, indivisible warehouse receipt — lives in the 1:1 lower NFT.
Missing link: a verifiable bridge between the registrar (B3/CERC, on Corda) and the public chain.
Giants Labs · Brazil warehouse-credit · instruments per CVM 175 · Lei 14.430/2022 · 15 / 28
16
The capital rail — each instrument to its own DeFi market
We don't run a lending market — we onboard collateral into the ones that already exist. Each instrument has a different risk profile, so a different on-chain market. The token layer lives offshore; disbursement to the farmer stays onshore, in reais.
Tokenized instrument
Matching DeFi market
WA · warrantlow risk
Pledge right (penhor)
lien on the grain over the CDA title · held by the senior lender · LTV 50–60% at launch
→
Aave
money market (RWA track, Horizon) · collateral for a USDC loan
CDA · warehouse titlemedium risk
Ownership token (title)
title to the grain lot · full price exposure, no LTV buffer
→
Morpho
isolated markets
Centrifuge
RWA marketplace
CPR · future cropnot at launch
Trade receivable
pre-harvest forward — the half of the cycle we don't refinance · mapped for later
→
Pendle
fixed yield
Maple
private-credit pool
funds the pools
Global USDC capital
investors pick the pool at their risk level
The rail: tokenize → place on the matching DeFi market by risk → global USDC capital funds it; our seat is the rail, never the credit book. Launch = the senior WA lane; CPR is mapped, not built. Needs a registrar (B3/CERC) → public-chain bridge (see “Two layers”). LatAm precedent: Credix (Solana) ran on-chain agro credit in Colombia from 2023, then moved to SME — Brazil's warehouse-credit rail is unbuilt.
Giants Labs · Brazil warehouse-credit · DeFi lending: Aave · Morpho · Centrifuge · Pendle · Maple · token layer offshore, disbursement onshore in BRL · 16 / 28
17
The price feed — public indices, an oracle to carry them on-chain
The price source is boring on purpose — grain settles daily against public benchmarks. But a lending protocol can't read a website: an oracle still has to deliver the numbers on-chain.
The feed, step by step
1 · Sources
CEPEA/ESALQ · B3 settlement · CBOT/ICE close — public references, correlated series counted once (B3's corn settle is built on the ESALQ index).
2 · Formula
Port benchmark − the route's measured freight — basis as a formula input, not an average · adjusted for the lot's grade.
3 · Delivery
Signed on-chain update — 24-h heartbeat + 2% deviation trigger. Daily cadence is enough for grain.
4 · Fail-safe
Stale feed ⇒ new borrowing pauses. The market fails closed, not open.
Three ways to run it — open choice, set by the protocol's listing rules
A · Oracle network
Chainlink-type custom data feed — what Aave listings effectively require; Morpho is oracle-agnostic.
B · First-party signed
The index provider or registrar signs its own prices — cheapest, fits a pilot.
C · Attestation + dispute
Optimistic push with a bond and challenge window — suits slow assets.
Bounded role: the oracle only triggers states — margin call, default. The realization price is set by the statutory exchange auction (art. 17 §2º, Lei 11.076), with the standby bid as its floor (p. 19 · structure to be confirmed with counsel) — never by the platform, and there is no manual override: stale sources ⇒ the market fails closed. Basis is a deduction, not noise: 12–20% of the port price in Mato Grosso, 3–4% in western Paraná — subtracted per route, never averaged.
Giants Labs · Brazil warehouse-credit · daily public benchmarks · oracle design A/B/C · 17 / 28
18
Default day — the mechanics LPs ask about first
Every serious capital conversation reaches the same question first: what exactly happens on default? The answer is a pre-arranged, extrajudicial path where the lender never touches the grain.
1.67×
collateral coverage at 60% LTV — collateral falls below the loan only after a >40% drop; via the standby-bid channel the pool stays whole to ~25% (p. 19; ~30% at a 15% discount).
12–20%
the measured route basis in Mato Grosso (3–4% in western Paraná) — deducted in the collateral formula up front, not left inside the buffer. Price truth arrives via the oracle layer (p. 17).
180+ days
the honest caveat: in judicial recovery, courts can freeze enforcement (essencialidade). Mitigated by insurance + off-take — not eliminated.
Pre-arranged is the keyword: the liquidation buyer, the insurance subrogation and the warehouse's obligations are contracted at issuance — default day executes a checklist, it doesn't start a negotiation. The buyer seat is a standby commitment — next slide. Sale mechanics follow the statutory auction path (Lei 11.076) — structure to be confirmed with counsel.
Giants Labs · Brazil warehouse-credit · extrajudicial enforcement per Lei 11.076 · mechanics & parameters to be confirmed with counsel · 18 / 28
19
The liquidator — a standby bid, not an auction hope
The first LP question: how fast does collateral become cash? Design: before any loan funds, a global grain desk signs a standby commitment: buy $1M+ seized lots within 24–72 hours at a 15–20% discount to the formula value. The lot changes hands on paper; the grain never leaves the elevator.
The math on a $1M lot · 65% LTV (the ceiling) · standby bid at 80%
What the desk will ask for — the validation agenda
Existence & quantity — continuous monitoring + independent audit trail
Grade — certificate at deposit, periodic re-inspection (CMA)
Clean title — CDA/WA registered at B3/CERC, no competing lien
Location — an elevator on a corridor the desk already runs
Legal path — pledge enforcement → extrajudicial sale, to be confirmed by counsel
Who holds this seat
desks that already buy Brazilian grain at scale — mapped targets, not commitments
What the 15–20% actually is: ~5–8pp title risk · 2–4 quality · 2–3 logistics & speed · 2–4 margin · ~0.5 financing — mostly title and quality, not 72-hour price risk, which is why every automated lien and title check compresses it. Status, honestly: this seat is the next roadshow — trader desks and shortfall insurers. The one question we bring: what has to be installed at the warehouse for you to sign?
Giants Labs · Brazil warehouse-credit · standby liquidation 24–72 h · 15–20% discount · targets, not commitments · 19 / 28
20
Stablecoins aren't banned — they're an FX operation
It's about regime, not prohibition. The central bank classifies USD-stablecoin flows as câmbio (foreign exchange) — legal, but only through an authorized institution. Three resolutions define the entire legal structure.
Res. 521 · câmbio
USD-stablecoin operations are foreign exchange — legal, but only via an authorized FX institution. $100k cap per operation with a non-authorized counterparty + reporting.
Res. 520 · art. 12
A VASP cannot extend credit. The lender of record must be licensed — SCD, bank or FIDC. The protocol is distribution, never the creditor — the same split our design makes (p. 7).
Res. 561 · 01 Oct 2026
eFX providers cannot settle the offshore leg with stablecoins — settlement must run through a formal câmbio operation. Precedent: Nomad / Braza Bank.
The compliant path — stablecoin as asset wrapper, not settlement rail
The loan can be dollar-denominated, but the money legs run onshore: the dollar enters via an authorized bank, the farmer receives and repays in BRL through the FIDC/SCD. Offshore token vault (e.g., a segregated-portfolio SPC) — to be confirmed with counsel.
The stack is honest: the investor's target return at the bottom, the tokenization layer's real costs on top, and the borrower still lands at roughly half the local BRL rate. No hedge line — a dollar loan against a dollar-priced crop.
The investor earns
8%
net target, in USD · senior · insured · over-collateralised · ~180-day paper
+
The tokenization layer costs
~3pp
origination & servicing · verification & monitoring · insurance · structuring — none of it is free
=
The borrower pays
≈11%
all-in, USD-denominated · disbursed & repaid in BRL onshore
vs
Local benchmark
18–25%
bank / barter rate in BRL — the rate the farmer faces today
Example: a $5M pilot pool at 50–60% LTV backs $8.3–10M of stored collateral · insurance covers 80–90% of a physical loss (price risk sits in the LTV buffer) · target borrowers: co-ops & traders, tickets from ~$1M. Why 8% clears: priced against passive on-chain dollar yields — insured senior at 8% is a large pickup. Currency: USD base case; the RMB leg — where yuan money fits and where it doesn't — is mapped on the next slide. Indicative, July 2026 — targets, not promises.
Giants Labs · Brazil warehouse-credit · USD-native · no FX hedge · RMB leg — next slide · indicative · 21 / 28
22
The RMB leg — where yuan money fits, and where it doesn't
China buys ~80% of Brazil's soy exports and ~41% of the bilateral trade already settles in RMB — so “why not cheap yuan funding?” is the first question from any Asian desk. Parity answers it: the 3% headline rate survives only while the money never converts. One door does that.
The parity trap — hedged RMB is just local money
~3% CNH + ~11.25pp forward points ≈ 14–15% BRL
Selic 14.25% minus CNH ~3% is the hedge cost — there is no direct BRL/CNY forward, only a synthetic USD cross. Hedged into reais, yuan funding lands exactly where Selic money already is. The low rate is real only against RMB revenue — RMB-settled soy at the exporter, never at the farm gate.
~4.5–5.5%
all-in RMB cost of the one clean channel — the export prepayment · 0% IOF · 0% withholding.
~41%
of China–Brazil trade settles in RMB (early 2025) — the plumbing exists: ICBC Brasil is the PBoC clearing bank, BOCOM BBM the first direct CIPS participant in South America.
Channel
All-in cost, 4-month tenor
Verdict
A · RMB export prepayment (PPE) — at the exporter/SPV
B · Chinese bank, onshore BRL — BOC Brasil rural credit, LCA-funded
est. ~12–16% BRL (LCA-implied; pricing not public) · no cross-border leg → no IOF, no WHT · a limited book, not the market-wide 18–25% — local BRL with a Chinese owner, not RMB capital
helps the farmer, not the cost
C · Cross-border CNH loan — Lei 4.131, sub-365-day
3.5% IOF on principal + 15% WHT + hedge ⇒ ~25% annualized on a 4-month tenor
fatal — IOF alone kills it
How it plugs in: the RMB prepayment funds the exporter/SPV leg against RMB-settled receivables — the farmer leg stays USD/BRL (p. 21). With the tokenization layer's ~3pp on top, the export-leg borrower lands at ≈8% all-in in RMB — the number beside the ≈11% USD. The 190bn-yuan PBoC–BCB swap line is a central-bank backstop, not lendable capital. RMB farm-level credit has zero precedent in Brazil; an RMB prepayment against tokenized receipts at the export leg would be a first — none found. Trigger to go deeper: the buyer's RMB-settled share >50% + a desk lending CNH +<300bp unhedged.
Giants Labs · Brazil warehouse-credit · RMB at the export leg, USD at the farm gate · parity as of July 2026 · 22 / 28
23
Tokenization already works — all of it whole-cycle
Brazil's RWA market has raised R$9.96 bn across 5,593 tokenized assets (RWA Monitor, Jul 2026) — and agro paper (CPR) is already ~13% of the tokenized stock. But look at what gets tokenized: whole-cycle credit paper. The stored-grain receipt — our lane — is a flat zero.
What's live on-chain — token stock by instrument, Jan 2026 (R$ bn)
RWA Monitor · Jan 2026 snapshot · tokenized CPR = ~0.07% of the R$560 bn CPR stock
Cumulative capital raised via tokenization (R$ bn · RWA Monitor)
~×24
tokenized-market volume in a year: R$0.12 bn (Feb '25) → R$2.88 bn (Jan '26)
5,593
tokenized assets live — the rail works; the lane is empty
The read: the market has proven it can tokenize agro credit — it just tokenizes the risky half. Nobody has printed an institutional track record on the warehouse-receipt rail; at ~0.07% penetration of the CPR stock, the race is for credibility, not share.
How our offer differs — the open lane in one table
The live agro-token players are real — and none of them does what this deck describes. They tokenize whole-cycle credit paper or run closed payment networks; nobody delivers the stored-grain receipt into open lending protocols.
Agrotoken
Rivool / Ecoagro
VERT
This project
What's tokenized
grain balances at partner silos — 1 token = 1 t (~$70M tx, 1,000+ farmers)
CRA/CPR credit notes (~$9.2M each live)
CRA securitizations — R$700M on XRPL
the stored-grain receipt itself (CDA/WA)
Cycle stage
post-delivery — but used as payments & barter
whole cycle — pre-harvest credit
whole cycle, wrapped for capital markets
post-harvest storage window only (~180 d)
Production risk reaches the investor
n/a — payments instrument
yes
yes, inside the wrapper
no — the asset already exists
Capital source
closed network + bank partnerships (BB / Broto)
on-chain note buyers · target 9–12%
institutional securitization desks
global lending protocols · investor 8% · LTV 50–60%
Exit on default
—
issuer / securitizer process
securitizer process
standby liquidator bid · 24–72 h · 15–20% discount
Closest peer: SiloReal (BR) digitizes agro assets including receipts — an asset-digitization play, not a lending rail. Argentina's Origino–Grassi runs a silobolsa-warrant NFT pilot (LTV up to 85% with a guarantee layer, results unpublished). The receipt-to-protocol lane is open — and it's the safest slice of the cycle.
Giants Labs · Brazil warehouse-credit · sources: Agrotoken · RWA.xyz 06.2026 · VERT · company disclosures · 24 / 28
25
The stakeholders — what each seat wants, and what we need from it
Four players touch the deal, each for a selfish reason — and three of the four open with the same question: is the grain really there? The platform is the shared answer. The design's job is to make the selfish choice also the one that keeps the grain verifiable.
Stakeholder
Who exactly
What they want from us
What we need from them
Borrowers
co-ops first; mid-size traders (cerealistas) next
cheaper money without selling the crop — and zero change to daily operations
grain in designated warehouses, title clean at pledge — any prior CPR settled at the gate from the draw · consent to independent monitoring · the first pilot lots
Warehouses
designated spaces with monitoring equipment
confirmations that pay — and not ending up the fall guy: ledger-vs-physics divergence surfaces in a journal, not in court
equipment install + data access · receipts issued on physical intake only · acceptance of an independent measurement channel
Liquidators
large trading desks that already buy at volume
a flow of discounted lots with known grade, clean title and a ready logistics file — no from-scratch due diligence
a pre-signed standby bid at “exchange clearing price − discount” · minimum-lot commitments · their warehouse install checklist, up front
Lenders
strategics first; institutional LPs
senior, insured exposure at a target 8% in USD (RMB — only via the export-prepayment leg, p. 22) · collateral they can verify line by line · default mechanics that execute
credit-committee requirements before the pilot, not after · a term sheet fixed at start · patience from pilot to scale
Everyone is selfish — by design. No seat can move the grain's state alone (p. 14), and every seat earns more when the record is verifiable. The named-candidate version of this map — ten seats, in deal order — is on p. 27.
Giants Labs · Brazil warehouse-credit · four seats · the shared asset is the verified grain lot · 25 / 28
26
What we don't know yet — said out loud
Research stage means open questions. These are the four we're resolving with partners — not hiding in an appendix.
Q1 · Origination channel
Co-ops, traders or brokers?
Co-ops — volume and warehouses, but closed data cultures. Traders / brokers — concentrated dealflow, export-native, easier data. Agri-fintechs — licensed and digital, but with their own funding agendas. Undecided — first conversations run all three in parallel; target borrowers: co-ops and traders, tickets from ~$1M.
Q2 · Free collateral
How much stored grain is actually free to pledge?
Most grain in store is already sold to traders or tied by pre-harvest CPR / barter — and the CDA/WA stock is so thin no regulator publishes it (slide 11). Mapping the truly unencumbered share, warehouse by warehouse, decides where the first deals source. Registry microdata needed — no desk estimate exists.
Q3 · Vault & wrapper
Where does the token live?
Onshore FIDC-quota tokenization (CVM 175) vs an offshore segregated-portfolio SPC holding the token layer. Cost, investor qualification and Res. 561 compliance pull in different directions — to be settled with counsel.
Q4 · Registrar → chain bridge
How does the title get on-chain, verifiably?
No verifiable B3/CERC → public-chain bridge exists yet. Build vs partner vs attestation-based interim — open. The price feed (p. 17) rides the same bridge. This is the core technical question for the tech-lead seat.
If you hold a strong view on any of these — that's exactly the conversation this deck is for.
Giants Labs · Brazil warehouse-credit · open questions, stated — not hidden · 26 / 28
27
The partner map — ten seats, in deal order
The build in the order a deal actually flows: who brings the borrower → who holds the grain → who proves it's there → who wraps it legally → who funds it → who sells it on default. Names are mapped candidates — live commitments are marked in the status column.
#
Role
What's needed
Mapped candidates
Status
1
Origination & dealflow
borrowers via co-ops / traders / brokers / agri-fintech
TerraMagna · Agrolend · Aegro
in talks
2
Warehouses & title issuance
licensed armazéns gerais issuing CDA/WA
co-op networks (Coamo · C.Vale-type)
mapped
3
Collateral verification
proof the grain exists: audits, satellite, traceability, collateral manager
The ask: take one of the ten seats on the partner map (p. 27) — or point me at the person who should hold it. One conversation is enough to start.
Where this stands · July 2026
Research stage, on the ground in Florianópolis (relocated long-term). Refinancing structure pressure-tested with impact investors & grain-storage practitioners · first soft LP interest — $2M, conditional on traceability · guest lectures at UFSC/LabSEC. Next: one discrete pilot deal, and the liquidator roadshow.
The fastest ways to help
01 · Intro
Originators & grain desks
a co-op, trader or warehouse network open to a pilot lot — or a desk open to a standby-liquidation talk
02 · Counsel
Legal
a recommendation into tier-1 counsel — câmbio, FIDC/CRA, VASP; the question map is ready to send
03 · Capital
8% senior
a first conversation with DFI / impact capital, Asia institutional desks, or an on-chain dollar pool
04 · Tech & verification
Build & trust layer
a senior builder for the contracts & registrar bridge — or monitoring platforms, CMAs and agro insurers for the trust layer
— open talks · tech-lead seat open
Nora (BRL-stablecoin · RWA distribution + the onshore BRL leg) · impact funds · on-chain pools