Report:

The Forrester Wave™: AI Infrastructure Solutions, Q1 2024

Top Vendor Products for AI Infrastructure Solutions in 2024

• Amazon Web Services
• Dell Technologies
• Google
• Microsoft
• NVIDIA

AI Infrastructure Solutions Vendor Product Comparisons 2024

• Amazon Web Services - Leaders
  

  Pros:

  • Solution ecosystem
  • Data preparation services
  • Inferencing
  • AI development tools

  Cons:

  • Could strengthen current offerings by adding more AI workload capabilities for on-premises and edge deployments
  • Declined to participate in the full Forrester Wave evaluation process


• Dell Technologies - Leaders
  

  Pros:

  • Architecture (with reference architectures)
  • Configurations
  • Model training

  Cons:

  • Could improve by designing AI-specific infrastructure management tools
  • Could include or partner more tightly with AI development tool providers
  • Has cloud capabilities for some AI workload components, but lacks a fully managed AI infrastructure solution


• Google - Leaders
  

  Pros:

  • Highest scores of all vendors in this evaluation across the board
  • Superior roadmap and innovation
  • Tight integration with Google's AI platform, Vertex AI
  • Efficient elastic scale to run spikey workloads and control costs

  Cons:

  • Customers don't have the option of using the totality of Google AI infrastructure on-premises


• Microsoft - Leaders
  

  Pros:

  • Architecture
  • Ecosystem
  • Data preparation
  • Model training
  • Inferencing
  • Development tools
  • Breadth of services for data preparation, application development, cognitive services, and pretrained models including OpenAI

  Cons:

  • Can improve by adding AI-specific infrastructure management tools
  • Could expand AI workload capabilities on Azure Arc for on-premises deployments


• NVIDIA - Leaders
  

  Pros:

  • System architecture
  • Partner ecosystem
  • Configuration options
  • Model training
  • System management
  • State-of-the-art systems designed by and directly available from NVIDIA

  Cons:

  • Relies on cloud service providers and other partners to accommodate scalable data preparation workloads


• Hewlett Packard Enterprise - Strong Performers
  

  Pros:

  • Configurations
  • Data preparation
  • Management software
  • Locale (cloud and on-premises)
  • Breadth of deployment options from edge to cloud

  Cons:

  • Can improve by expanding solution ecosystem
  • Needs to further optimize model training
  • Must expand messaging to growing numbers of enterprise AI developers
  • Needs to improve roadmap with AI-specific optimizations beyond adding next generation GPUs
  • Declined to participate in the full Forrester Wave evaluation process


• IBM - Strong Performers
  

  Pros:

  • Solution ecosystem
  • Configuration
  • Data preparation workloads
  • Locale (cloud and on-premises)
  • AI development tools
  • Option to run on-premises and/or cloud
  • Ability to leverage existing investments in IBM Storage and IBM Z

  Cons:

  • Can improve solution by adding more training workload optimizations
  • Needs to unify management tools for AI workloads
  • Cloud is up against formidable competition from three major public clouds
  • Needs to improve roadmap by better unifying how customers buy and consume AI infrastructure seamlessly across ecosystem


• Alibaba Cloud - Contenders
  

  Pros:

  • Data preparation
  • Management tools
  • Development tools
  • Fault-tolerance
  • Efficiency
  • Ability to scale AI infrastructure to handle huge AI workloads

  Cons:

  • Can improve by investing in more optimizations for model training and inferencing
  • Must expand beyond APAC region to be attractive to global enterprises
  • Needs to expand solution ecosystem
  • Vision, innovation, and roadmap can improve if more informed by peer hyperscalers


• Lenovo - Contenders
  

  Pros:

  • Solution ecosystem
  • Configuration options
  • On-par scores for model training, data, inferencing, fault-tolerance, and efficiency
  • Ability to run AI workloads on-premises and at scale

  Cons:

  • Could improve by offering an AI-specific management tool
  • Needs further optimizations for AI workloads
  • Lacks a superlative 'why Lenovo?'
  • Could improve innovation by partnering with AI/ML platform vendors offering AI full-lifecycle tools


• Oracle - Contenders
  

  Pros:

  • Data preparation
  • On-par scores for model training, inferencing, fault-tolerance, and efficiency
  • Cost competitive AI infrastructure option compared to other cloud providers

  Cons:

  • Could improve by adding further optimizations for AI workloads
  • Can improve strategy by greatly expanding AI infrastructure vision beyond raw GPU instances
  • Can improve roadmap with enhancements to AI-specific development tools


• Cerebras Systems - Challengers
  

  Pros:

  • Training workloads
  • Workload efficiency
  • Locale (on-premises and cloud)
  • Training performance

  Cons:

  • Could improve by expanding type of training workloads supported
  • Needs to fill gaps with more robust solution ecosystem
  • Generally needs to improve across the board to become more on par with other vendors
  • Roadmap relies on partners for data preparation and inferencing workloads


• Graphcore - Challengers
  

  Pros:

  • On par for model training, inferencing, development tools, efficiency, and locale
  • Price/performance delivery

  Cons:

  • Can improve by developing more sophisticated management tools
  • Needs fault-tolerant capabilities
  • Must expand solution partner ecosystem to fill capability gaps
  • Needs to expand vision, innovation, and roadmap to compete with larger players
  • Can accelerate go-to-market strategy through partnerships or investing more in sales and marketing

Reference

• Forrester, The Forrester Wave™: AI Infrastructure Solutions, Q1 2024, 17-Mar-2024, ID RES180430

View Market Definition
View Vendor Movements