This blog explores the strategic role of data center innovation labs in accelerating AI inference adoption, with insights from Renato Pasquini, Research Vice President, and Carina Gonçalves, Industry Principal, from Frost & Sullivan’s Data Center Services team.


 

As artificial intelligence (AI) moves from model training to production inference, organizations face a new challenge: deciding where AI workloads should run and how to validate complete AI solutions before deployment. This creates a strategic opportunity for retail colocation and interconnection providers to play an active role in AI adoption, by bringing together the partners required to deploy AI successfully.

As inference becomes embedded in business applications, infrastructure placement becomes more consequential. Some workloads will benefit from centralized scale; others will require proximity to users, enterprise data, or specific jurisdictions. Distributed inference therefore creates an integration challenge.

Unlike large-scale model training, inference workloads must often support real-time interactions, comply with data sovereignty requirements, and integrate directly with enterprise applications, making deployment location a critical design decision.

In this scenario, innovation labs can become an important mechanism for addressing this challenge, provided they connect technical experimentation with a credible path to production.

Operators are approaching this opportunity from different starting points:

 

Operator and initiative Demonstrated focus Implication for ecosystem development
Digital Realty — Digital Realty Innovation Lab (DRIL) Production-grade testing of multi-vendor AI, high-performance computing (HPC), and hybrid cloud architectures. There are five labs globally, located in Northern Virginia, London, Tokyo, São Paulo, and Singapore, with the latter two launched in November 2026. Creates a shared environment for validating how infrastructure components and partner solutions work together.
Equinix — Enhanced Performance Innovation Centers (EPIC) Testing liquid-cooled graphics processing unit (GPU) and central processing unit (CPU) platforms inside operational data centers, building on earlier co-innovation facilities and Solution Validation Centers. There are 5 operational centers globally: London, Paris, Frankfurt, Osaka, and Sydney. Helps customers and technology partners resolve deployment and operational questions before scaling.
ST Telemedia Global Data Centres and SuperX — Singapore AI Innovation Centre Launched in April 2026 at STT Singapore 5, offering GPU infrastructure for pilots, proofs of concept and model benchmarking, with options to progress into production. Extends the lab proposition toward accessible AI compute and workload experimentation.
NTT DATA — Data Center Trial Field, Japan Established in November 2024 to support joint verification of liquid-cooling technologies and address differences between manufacturers’ specifications and operating practices. Illustrates how shared testing can improve interoperability and operational readiness across suppliers.

 

Collectively, these initiatives demonstrate growing industry recognition that AI adoption requires more than additional power and cooling capacity. However, their objectives vary considerably, ranging from infrastructure validation to broader ecosystem enablement.

“The strategic value of an innovation lab lies in helping an ecosystem deliver together. As AI inference becomes more distributed, enterprises need confidence that their AI solutions will perform reliably in production. Colocation operators can bring those partners together, validate deployments, and shorten the path from experimentation to production.”
— Renato Pasquini, Research VP & Global Program Leader, Frost & Sullivan

In November 2026, Carina Goncalves attended the launch of Digital Realty’s DRIL in São Paulo, operated through Ascenty. She noted that the lab already brings together 10 direct partners, five indirect partners, and another 10 certified business partners with installed racks. The breadth of participation is significant because enterprises increasingly require integrated solutions spanning infrastructure, connectivity, cloud access, software platforms, and professional services. Bringing these capabilities together in a single environment reduces deployment risk and can accelerate AI adoption.

“Ascenty’s innovation lab offers a real data center environment to test and validate technologies such as liquid cooling, multicloud interconnection architectures, and AI-optimized computing. This ecosystem is essential for evaluating interoperability among servers, storage, networks, and software platforms.”
— Carina Goncalves, Industry Principal, Frost & Sullivan

Bringing these partners and capabilities into one environment creates an opportunity to evaluate complete architectures against local enterprise requirements. For customers, the potential value is greater confidence in technology selection, integration, and deployment costs. The next measure of success will be how effectively this collaboration produces repeatable solutions that move from lab validation into production.

In sectors such as government, financial services, healthcare, and critical infrastructure, innovation labs can also help validate architectures designed to satisfy sovereignty, residency, and compliance requirements before production deployment.

Historically, innovation laboratories focused on validating individual infrastructure components. AI inference introduces a different requirement: validating complete business solutions composed of infrastructure, models, data, software platforms, security controls, and operational processes. The operators that facilitate this broader validation process may become increasingly influential participants in enterprise AI decision-making.

From Infrastructure Readiness to Solution Readiness

At the infrastructure level, labs should help customers evaluate power density, cooling, GPU utilization, storage performance, and network behavior under representative workloads. For inference, evaluation should extend to: time-to-first token, response latency under concurrency, resilience, and cost per workload. Testing alternative deployment locations can reveal whether proximity benefits justify the cost of distributing capacity.

At the solutions level, the opportunity is to assemble GPU-as-a-Service providers, model platforms, software vendors, systems integrators, and managed service providers around customer use cases. An enterprise knowledge assistant, for example, requires more than GPU capacity: it needs secure access to enterprise information, retrieval, model serving, monitoring, and application integration.

The lab can provide a setting where these partners jointly validate the solution and establish who will deploy, support, and operate it.

Equinix’s inaugural Horizon customer and partner event on September 2, 2026, illustrates this broader ecosystem direction. Its announced Inference Exchange combines Equinix infrastructure and connectivity with NVIDIA reference architectures and Together AI’s inference platform, with availability scheduled for Q1 2027. This is separate from EPIC, but it shows how infrastructure validation and partner-led services can reinforce each other.

A Commercial Discipline for Ecosystem Development

For retail colocation operators, the strategic logic is compelling: helping customers validate deployments can create demand for capacity, interconnection, managed infrastructure, cloud connectivity, and partner-led services. The operator can coordinate this process while maintaining clear boundaries with partners that supply compute platforms, applications, and integration.

Success should be measured through pilot-to-production conversion, deployment time, repeatable architectures, and incremental ecosystem business. Events can initiate relationships; labs should turn those relationships into validated deployments.

As AI inference expands, the strongest value proposition will combine AI-ready infrastructure, ecosystem partners, and validated evidence that a customer’s workload can perform reliably, securely, and economically in production. Innovation labs can help operators bring those elements together and make ecosystem development a measurable part of their growth strategy.

Strategic Priorities Shaping Data Center Growth

Explore Frost & Sullivan’s in-depth, executive analyses on data center colocation, infrastructure investment priorities, and emerging growth opportunities:

  1. Frost Radar™: Data Center Colocation Services in North America, 2026
    Explore the competitive landscape of North American data center colocation providers through Frost & Sullivan’s growth and innovation benchmarking.
  2. Top 10 Growth Opportunities in Data Center Services, 2026
    Examine growth opportunities shaping data center services, including infrastructure optimization and evolving customer requirements.
  3. Data Center Colocation: Decision-makers’ Insights and Strategic Priorities, 2026–2027
    Understand enterprise colocation priorities and the factors influencing infrastructure and service provider decisions.

Ready to Lead the Transformation?

About Renato Pasquini

Renato Pasquini has been working for +19 years with consulting and research in the information and communication technologies (ICT) sector, and is currently research vice president at Frost & Sullivan. Pasquini also serves as an advisor to organizations and associations in Latin America. Pasquini holds an undergraduate and master's degree in Business Administration from EAESP-FGV, a postgraduate degree in Economic Law from GVLaw and a masters in ICT Business from UPC-Spain.

Renato Pasquini

Renato Pasquini has been working for +19 years with consulting and research in the information and communication technologies (ICT) sector, and is currently research vice president at Frost & Sullivan. Pasquini also serves as an advisor to organizations and associations in Latin America. Pasquini holds an undergraduate and master's degree in Business Administration from EAESP-FGV, a postgraduate degree in Economic Law from GVLaw and a masters in ICT Business from UPC-Spain.

Your Transformational Growth Journey Starts Here

Share This