36-Month Action Plan
From zero to frontier AI lab. Phase by phase, dollar by dollar, milestone by milestone.
Phase 0: Foundation (Months 1-3)
Budget: $500KGoal: Incorporate, build founding team, release first model, establish government relationships.
Register in Uruguay (stable legal framework, Mercosur treaty access) or Delaware + Argentine subsidiary. Uruguay offers political stability and treaty access.
5-7 founding members: 2 ML researchers, 2 engineers, 1 product, 1 government relations, 1 CEO/founder. Target Argentine/Brazilian talent abroad.
Secure $500K-1M from angels, regional family offices, government grants, and sovereign wealth funds.
Fine-tune and release "Plata 7B" - Spanish and Portuguese optimized, Apache 2.0 license.
Build simple inference-as-a-service platform with Spanish/Portuguese endpoints.
Establish advisory board with ex-ministers and officials from each target country.
Pitch to Argentine AI czar, Brazilian digital transformation office, Chile CORFO.
Apply for Argentina Innovadora, BNDES innovation, CORFO Chile, and Itaipu tech fund.
Set up development hub in Buenos Aires or Montevideo (talent + cost + stability).
Begin feasibility discussions for data center partnership at Itaipu.
Build website, positioning, and community (Discord/Telegram for regional AI developers).
Phase 1: Product-Market Fit (Months 4-12)
Budget: $3-5MGoal: First paying customers, second model release, Series A preparation.
Close 3-5 government pilot contracts at $100K+ each. Target: Uruguay digital government, Argentine tax office, Chile Codelco.
Land 5-10 enterprise API customers. Target: regional banks, telecoms, and fintechs.
Release MoE model (8x7B equivalent) optimized for legal/financial domains.
Launch consumer/business chatbot (Spanish/Portuguese first, English second).
Hire to 20-30 people. Focus on engineering and government sales.
Raise $3-5M seed. Target regional family offices, a16z/Lightspeed LatAm interest, sovereign funds.
Open small GPU cluster (32-64 GPUs) at Itaipu for training and inference.
First government on-premise deployment (no internet connection, full sovereignty).
Launch public-facing AI program (citizen services, job matching, legal aid).
Publish 2-3 papers at NeurIPS/ICML/ACL to establish technical credibility.
Establish partnerships with UBA, USP, UChile for talent pipeline and research.
Phase 2: Scale (Months 13-24)
Budget: $20-50MGoal: Scale to 100+ employees, Series A, first pre-training run, multi-country deployment.
Raise $20-50M Series A. Mix of regional and US/European VCs.
Grow to 80-100 people. Open second office (São Paulo or Mexico City).
First from-scratch pre-training run (50B+ parameters). Major technical milestone.
Launch agent builder for government and enterprise workflows.
Launch custom model training platform for enterprise fine-tuning.
Enter Mexico, Colombia, and Peru markets.
Achieve $5M+ ARR from government and enterprise contracts.
Multi-year, multi-country government framework agreements.
Build large-scale training facility at Itaipu (512+ GPUs).
Launch sovereign cloud offering (government-only, air-gapped, fully compliant).
Evaluate acquisition of regional AI startup or data center operator.
Phase 3: Frontier (Months 25-36)
Budget: $100M+Goal: Undisputed regional AI leader, Series C, international expansion.
Raise $100M+ for frontier model training and international expansion.
200-300 employees. Regional talent hub.
100B+ parameter model competitive with GPT-4 class. The "Plata Grande" moment.
Launch in Spain and Portugal. Reverse language market expansion.
Achieve $20M+ ARR from diversified revenue streams.
Begin IPO preparation or strategic acquisition discussions.
Establish Plata AI as the "Mistral of Latin America" in global perception.
Risk Register
| Risk | Likelihood | Impact | Mitigation |
|---|---|---|---|
| Funding gap | Medium | Critical | Bootstrap with consulting; government grants; phased approach |
| Talent drain | High | High | Above-market pay; heavy equity; remote work; US/EU rotation |
| Compute cost | High | High | Itaipu cheap power; cloud credits; NVIDIA startup program |
| US competition | High | Medium | Sovereignty angle; local data; government preference |
| Political change | Medium | High | Multi-country strategy; non-partisan positioning |