Bridging the Computing Power Divide: Building a New Digital Engine for Central America

BSIC & Tigo El Salvador Strategic Cooperation Project

According to McKinsey and Goldman Sachs, global demand for computing power is surging due to explosive growth in AI, large-scale model training, and smart terminals. Data centers have become the cornerstone of the digital economy.

Global Trend

USD 1T data center market by 2034

Regional Need

66% lack quality 4G/5G coverage

Policy Support

Tax exemptions & government incentives

Opportunity for Partnership

Aligns with global investment trends and local needs

Six Data Centers to Build a Regional AI Computing Power Hub

Project Scale

Construction of 6 high-performance AI computing centers in the capital and surrounding areas, capable of supporting millions of concurrent tasks and petabyte
-scale data operations.

  • Proprietary GPU scheduling platform
  • In-house modeling hardware
  • Energy-efficient green data center design

Dual Purposes

Internal Use

For BSIC’s own AI model training and service deployment.

External Services

Providing computing power to support Tigo’s operations and collaborative initiatives with Central American universities and government services.

Technical Highlights

Featuring proprietary GPU scheduling platform, in-house modeling hardware, distributed deployment, and energy-efficient green data center design.

Empowering BSIC's AI Strategy

Core R&D directions leveraging the high-performance computing infrastructure

Digital Human Generation

Develop realistic AI avatars for Spanish, Portuguese, and other languages to enable global digital interaction.

Voice Cloning

Use voiceprint modeling to replicate speech characteristics for immersive content delivery and cross-language communication.

3D Portrait Modeling

Leverage in-house scanners with parallel capture to enable real-time 3D modeling for various industries.

AI Recommendation

Integrate visual, audio, and text data to build intelligent recommendation systems and interest graphs.

Edge AI Compression

Perform pruning and quantization for edge models to optimize local AI response with minimal power consumption.

Web3 Digital Identity

Explore integration of virtual identity and social behavior in decentralized Web3 applications.

End-to-End In-House Development

Building a High-Performance and Reliable AI Computing Platform

Proprietary Hardware

  • HMC-VII scanners with 0.1mm precision
  • GPU-dense servers with A100/H100 accelerators
  • Industrial-grade projection-camera arrays

Software Platform

  • GPU task scheduling system
  • Compatible with PyTorch, TensorFlow, ONNX
  • Edge-cloud collaborative architecture

Green Design

  • PUE below 1.3 with liquid/air cooling
  • Integration with renewable energy
  • Sustainable computing practices

Three-Phase Deployment Roadmap

Building a Distributed Regional Computing Network

Phase 1: Pilot Deployment (2025)

Foundation

  • Launch the first data center in San Salvador
  • Integrate with Tigo HQ systems and local service calibration
  • Launch pilot services including AI customer service and university AI training

Phase 2: Regional Expansion (2026)

Growth

  • Build and connect the remaining five data centers
  • Expand customer base in academia, public sector, and tech parks
  • Build local brand presence and partner networks

Phase 3: Ecosystem Integration (2027)

Maturity

  • Establish developer portal and API marketplace
  • Promote regional interconnection with neighborin countries
  • Organize AI application summits and educational programs