At VMworld last year, VMware and NVIDIA announced a partnership to unleash AI for every enterprise. We’re excited to announce today that VMware vSphere 7 Update 2, in combination with NVIDIA AI Enterprise software suite delivers the next step of that vision. We’ll look at ‘what’ we’re announcing and ‘how’ it helps our customers – but first we’ll look at ‘why’.
Enterprises today look to AI to transform their business. They are looking at how to best invest in AI workloads such as training, inferencing, as well as data analytics…and they’re doing this because they believe it will deliver valuable business outcomes. According to a Gartner survey which asked about the rationale for investing in AI vs other technology options – the top 3 reasons include revenue increase or cost reductions, the need to stay competitive, and faster funding approvals.
While enterprises see tremendous value in AI, there are also significant challenges and barriers. The Gartner survey also found that many projects never make it from pilot to production, and top barriers to AI implementation included complexity of infrastructure integration, as well as security and privacy.
There are 3 key areas enterprise IT decision-makers should consider when investing in new AI projects to ensure they are successful:
- Risk: How do I manage the risk of pulling together solutions from disparate AI frameworks, enterprise software, and hardware components that can often be incompatible? Will it work well?
- Performance: How do I ensure the compute-heavy requirements for AI workloads are met? How can I ensure fast time to market by running projects and training models quickly?
- Scaling: How can I eliminate shadow AI and scale effectively in production? How can I make efficient use of precious GPU resources while ensuring enterprise manageability and availability?
The AI-Ready Enterprise Platform
VMware vSphere 7 Update 2 delivers powerful new support for the latest NVIDIA Ampere-based A100 GPUs, including enhancements to performance boosting GPUDirect communications and new multi-instance GPU (MIG) partitioning. vSphere also delivers the enterprise virtualization goodness for GPU powered AI workloads that our enterprise customers have come to depend on for CPU powered workloads – vSphere is the only virtualization platform that enables vMotion for NVIDIA vGPU powered VMs – simplifying infrastructure maintenance while also powering VMware DRS for automatic initial workload placement for AI infrastructure at scale.
NVIDIA AI Enterprise software suite is an end-to-end cloud-native suite of AI tools and frameworks, optimized and exclusively certified by NVIDIA to run on VMware vSphere. This includes frameworks that are broadly applicable and used across vertical industries such as manufacturing, logistics, financial services, retail, and healthcare. For example, it includes:
- TensorFlow and PyTorch for machine learning
- NVIDIA Tensor RT, for GPU optimized deep learning inference and NVIDIA Triton Inference Server to deploy trained AI models at scale
- RAPIDS, for end-to-end data science and analytics pipelines
This unique combination of the NVIDIA AI Enterprise software suite with VMware vSphere is the next step in delivering a powerful AI-Ready Enterprise Platform.
“With VMware and NVIDIA’s partnership, organizations can finally extend the benefits of a common enterprise infrastructure for their AI initiatives”. A single platform for consolidating AI and Enterprise workloads provides better security, manageability, and resiliency for their AI workloads. Customers can now accelerate their AI initiatives with confidence, knowing that such initiatives will scale well.”
– Ashish Nadkarni, Group Vice President, IDC
With the AI-Ready Enterprise Platform, our customers will be able to:
- Deploy with confidence: Customers can now deploy an end-to-end certified AI solution that they know will work, and that is supported by NVIDIA. This also means that they can avoid running AI projects in AI silos, by folding them into their existing enterprise infrastructure. Data scientists get to work with ready-to-go frameworks that they need to get started quickly, while IT operators get to leverage their existing enterprise platforms to ensure adequate governance and security.
- Boost AI performance: vSphere support for A100 GPUs delivers world-class AI performance – up to 20X the performance of previous-generation GPUs. As well, customers get near bare-metal performance while technologies such as GPU Direct communications also enable higher performance for scale-out workloads. For a deeper dive on performance, see this NVIDIA blog.
- Scale without compromise: With NVIDIA Virtual GPUs (vGPUs) enabling both time-sliced sharing as well as new multi-instance GPU (MIG) hardware-level spatial partitioning, customers have multiple options in how they can share valuable GPU resources across more data scientists and increase GPU utilization as they scale AI workloads in the enterprise. vSphere DRS automatically places workloads across AI infrastructure at scale for optimal consumption, and vSphere vMotion provides live migration to simplify infrastructure maintenance such as consolidation, expansion, or upgrades.
“For enterprises to stay competitive, it’s no longer a question of whether they should modernize their infrastructure for AI – it’s a question of how quickly they can do so while managing risk. The AI-Ready Enterprise platform from VMware and NVIDIA gives our customers the confidence to get started today.”
– Krish Prasad, SVP and GM, VMware
We’re excited to deliver this next step in unleashing AI for every enterprise and look forward to continuing the journey to deliver AI innovation to our customers in the future as well!