It was only in 2023 when generative AI (GenAI) exploded onto the market, changing the way organisations can monitor environments, create content, and automate operations. Organisations saw great potential, especially in AI applications improving customer support experiences, or creating content for sales, marketing, and other functions. Yet organisations also saw the risk of sharing their trusted data with cloud-based GenAI tools because that data could be used for training other AI models that competitors may access. While the focus has been on large language models (LLMs), most organisations only require AI models that are fine-tuned and customised with their domain-specific data to meet their unique business needs.
Our Asia-Pacific customers want to understand how they can gain the benefits of GenAI without sacrificing privacy or control of their data. Below are answers to the five questions we most frequently hear from our customers:
1. We want to embrace GenAI tools and applications, but training LLMs is expensive and time consuming. What’s the best approach?
One of the myths of GenAI is that you have to build and then train LLMs from scratch. That can be expensive and time consuming. However, you can start with a good foundation model (open source or commercial) and fine-tune it with your private business data. This is the lowest cost and best approach to take. The key to this is to take an architectural approach that is open and flexible. This allows you to experiment with different AI models and fine-tune them with your domain-specific data. In this case, you bring your AI model adjacent to your data, so your data never leaves your data center. It remains within your control so you maintain privacy and compliance. We call that Private AI.
2. We have specific compliance and security requirements for our data and AI applications. How do I ensure that we maintain security and compliance?
The best way to ensure security and compliance is to consider a Private AI architecture. There are three core principles to Private AI:
Highly Distributed: compute capacity and trained AI models reside adjacent to where data is created, processed, and/or consumed, whether the data resides in a public cloud, private cloud, data center, or the edge. Organisations keep control of their data and AI models, maximizing security and privacy.
Data Privacy and Control: an organisation’s data remains private to the organisation and is not used to train, tune, or augment any public models without the organisation’s consent. The organisation maintains full control of its data without taking on the added risk of data leakage.
Access Control and Auditability: access controls are in place to govern access and changes to AI models, associated training data, and applications. This allows organisations to showcase that they are implementing GenAI in accordance with policies and regulations around the responsible, ethical, and unbiased use of AI. Because the AI model and associated services run adjacent to your enterprise data, you may also benefit by using existing controls, tools, and processes to manage access and audit.
3. AI technology is moving so fast. What if I invest big in a specific AI technology, and then a new one comes to market that would work better for us?
The AI landscape is moving very fast and new technologies, models, services, and tools come to market all the time. The best approach is to leverage an AI infrastructure that is open, flexible—not tied to any single AI model or hardware stack—and is supported by an extensive ecosystem of industry-leading AI vendors. With maximum flexibility to work with a wide range of AI technologies, you won’t get locked into one particular solution. You can change models, services, and tools with a simple software update. VMware Private AI is that architectural approach. It delivers a range of GenAI solutions for your environment—NVIDIA, Intel, IBM, Hugging Face, open–source community repositories, and independent software vendors. And it is backed by an extensive open ecosystem: most major server OEMs (Dell, Lenovo, HPE, Supermicro, and more) provide certified server architecture. Industry-leading global system integrators like NTT Data, WWT, HCL, Wipro, Kyndryl, etc. can help customers along their GenAI journey.
4. AI experts are telling us that GenAI is beyond our reach because we need a massive number of GPUs to get started. We don’t have the necessary processing power, technical skills and budget so our only option seems to be that we run all our services with public cloud providers. Is that true?
No, it is not true. It is one of the top myths about GenAI. We hear from our Asia-Pacific customers that they are told to effectively leverage AI for their organisation that they need deep expertise, tens of millions of dollars, and thousands of GPUs. Not true. Most organizations can benefit from AI by running models fine-tuned with their data. They do not have to take on the cost burden of model training to benefit from AI. You can download a very good foundation model, even an open-source model, and then use that model with a technique such as retrieval augmented generation (RAG) connected to your secured data sources. Bringing a model to your data would allow you to control of your data and whether it should be used to benefit another model. You’re maintaining complete control of it, with privacy and compliance. With this approach, you could have a proof of concept (POC) up and running in weeks.
5. I have an existing VMware Cloud Foundation (VCF) environment. How can I make the most of my current investment while leveraging GenAI for our organisation?
If you already have an existing VCF environment, it is very easy to leverage your current investment to deliver GenAI capabilities to your customers and employees across Asia-Pacific. VCF is the central component in VMware Private AI. While allowing you to bring AI models to the data sources you already have, VMware Private AI maintains privacy, governance, and controls that are already in place—using your existing toolset.
ACCELERATE AI ADOPTION SECURELY
When it comes to AI, there is no longer any reason to debate trade-offs in choice, privacy, and control. Private AI empowers organisations with all three, enabling them to accelerate AI adoption while futureproofing their AI infrastructure.