AI, Data Centers and Digital Infrastructure: Ciena's Outlook for 2026
AI demand is reshaping how Latin America funds terrestrial networks, submarine cables, data centers and optical transport. Ciena's study on connectivity in the AI era points to one takeaway: data center interconnect is moving center stage, and high-capacity wavelengths (≥800 Gb/s up to 1.6 Tb/s) are becoming table stakes to move traffic between distributed AI nodes.
What the data signals
Training and inference place different pressures on the network. Training pulls massive volumes across long-haul links, while inference wants to live near users for latency. According to Fernando Capella, Ciena's Brazil country manager, hyperscalers are committing to the region with activity in Brazil, Mexico, Chile, Colombia and Argentina.
Where capacity must grow
Interconnection between data centers will define performance and cost for AI workloads. Expect upgrades to 800G-1.6T per wavelength as a requirement, not a nice-to-have.
Ciena brought 1.6 Tb/s per channel to market via its WaveLogic 6 family, signaling where long-haul and DCI designs are headed. For context on Ciena's portfolio, see Ciena.
Subsea: modernize first, then build new
Latin America already has an established subsea footprint. While cables are often modeled with ~25 years of useful life, that's more economics than physics. New optics can pull more capacity from existing wet plants, extending revenue potential without laying new cable.
New builds are still coming, but there's a queue. Cable production, deployment windows and ship availability are tight, which means lead times you need to plan around.
Hotspots for data centers and interconnect
- Brazil: more than 40% of LatAm data centers; primary demand driver.
- Mexico: Querétaro is a major cluster and adds geographic diversity relative to US facilities.
- Chile: strategic on the Pacific side due to cable landings.
- Colombia and Argentina: emerging footprints; OpenAI has signaled interest in Argentina.
- Caribbean: activity tied to subsea projects.
- Paraguay: interest tied to power availability from Itaipu for data center and mining plays.
Beyond DCs: 5G and broadband keep pushing traffic
Data centers lead this growth wave, but they're not alone. 5G rollouts and fiber expansion continue across the region. Brazil is ahead on fiber penetration; other countries are scaling up, adding steady baseline growth.
What providers are buying
Two threads stand out. First, high-performance optical transport for long-haul terrestrial and subsea routes. Second, pluggable optics that snap into routers (both Ciena and third-party) to scale metro and aggregation networks-ideal for ISP footprints.
In Brazil, ISPs are under heavier competitive pressure. That forces network design to double as a product strategy: better experience per dollar, smarter capex, and operational efficiency.
2025 performance and the competitive picture
Ciena reports strong demand through its fiscal year ending October, bolstered by optical leadership and alignment with AI infrastructure needs. Capella notes market validation as well: over the last 12 months, Ciena's share price rose 166%.
2026: practical headwinds and steady demand
Election years can slow decision cycles, but network demand tends to push through. Traffic has historically grown 20-30% per year; AI is adding another layer on top of that. Plan budgets and lead times accordingly.
What IT and Development leaders should do now
- Set a DCI upgrade path targeting 800G-1.6T wavelengths where traffic warrants it.
- Shorten time-to-capacity with pluggables (ZR/ZR+ class) in metro and aggregation layers.
- Place inference closer to users; align edge sites with peering, power and cooling realities.
- Model multi-year traffic with two curves: base growth (20-30%) and AI-driven surges.
- Expect long lead times for subsea and long-haul builds; lock vendor and ship windows early.
- Design for modular upgrades: coherent optics, flexible-grid ROADMs, and router line cards that won't bottleneck optics.
- Automate telemetry and capacity triggers to avoid manual firefighting during AI peaks.
- Run interop tests across vendors to avoid surprises in mixed environments.
Level up your AI skills
If your team is building or supporting AI services, a focused upskilling track helps you move faster. Explore practical certifications for engineers at Complete AI Training.
Your membership also unlocks:
