The Gen AI Healthcare Ecosystem Around Big Tech

Publication Date: 13/10/2023

Cranfield, UK, 13th October 2023 – In the healthcare industry, there is a lot of talk about the potential benefits of generative AI solutions. However, it is important to distinguish between what is real and what is just hype.

To better understand the practical applications of this technology, many healthcare providers, vendors, and consumers are asking important questions such as: What are the realistic use cases? How soon can we expect to see new developments? What are the factors that drive or hinder progress? What is the competition doing, and what strategies are they using? To answer these queries and more, I am launching a series of Insights aimed at helping you navigate this rapidly evolving field. This is a precursor to more in-depth analysis provided in Signify Research’s Generative AI Market Intelligence Service.

In this initial installment, I aim to provide a solid foundation for those unfamiliar with the complexities of gen AI. This Insight examines the gen AI value chain depicted in the graphic below, starting from the bottom at the hardware segment of the value chain, moving up, and covering all the stages before gen AI products reach their end user.

You can download a copy of this insight or simply continue reading.

Figure 1: Generative AI Value Chain

This Insight compares the Big Tech organisations along the gen AI value chain to support the analysis. In the context of this Insight, Big Tech refers to the three most significant and most advanced cloud providers: Google Cloud Platform (GCP), referred to as Google, Amazon Web Services (AWS), andMicrosoft Azure.

It’s important to note that this isn’t a technical comparison of the AI offerings provided by Big Tech. Instead, it is a testament to their significance in the future of the gen AI ecosystem, based on their strong foothold today.

Started from the Bottom

Gen AI is a remarkable feat of technological innovation. By utilising the potential of deep learning and neural networks, this groundbreaking technology opens a world of possibilities, enabling computers to generate text, images, and even audio.

Gen AI models differ from traditional machine learning (ML) models in their approach to learning. While traditional ML models rely on supervised learning with labelled data, gen AI models use unsupervised learning on large amounts of unstructured data, as depicted in the graphic below.

Figure 2: Traditional Machine Learning Models vs Foundation Models

Foundation models are very complex and require considerable computational power for training and inference. That is why gen AI models need specialised computing.


Specialised computer hardware is often used to execute ML and AI programs faster and with less energy, reducing the underlying operating costs. The most common type is graphics processing units (GPUs) clustered together to form a supercomputer. That is how OpenAI’s ChatGPT model was trained. NVIDIA is the market leader in the GPU segment, with analysts estimating over 90% of the world’s market share. All Big Tech providers utilise NVIDIA’s processing technology for ML.

However, some Big Tech players are also in the hardware market race. AWS has its Inferentia and Trainium chips designed to support heavy workloads generated from training and deploying gen AI models. Google has also been utilising its hardware infrastructure to support neural network machine learning in the Tensor Processing Unit (TPU) since 2015.

On the other hand, Microsoft Azure has not yet developed its own solution but has the world’s largest cloud investment in field-programmable gate arrays (FPGAs) using Intel solutions. Recently, it has been reported that OpenAI is exploring making its own AI chips, seemingly entering the race.


In addition to hardware, software is required to facilitate the development of FMs. The sentiment surrounding cloud technology is positive, with most healthcare providers eager to embrace its potential benefits. Advocates argue that scaling is easier, allowing hospitals to redirect resources towards improving patient outcomes, more cost-effective, and the deployment is faster than traditional on-premise servers.

Big Tech cloud offerings overlap significantly; hence, it depends on businesses’ specific requirements for which cloud provider to choose. Some providers use a multi-cloud approach to avoid being locked in with one vendor and be better prepared for disaster recovery.

All three Big Tech firms are big players in healthcare cloud provision. Among these, Azure and AWS are the leading platforms used by more than half of all healthcare providers. Google is also used, but it has a smaller market share. The other prominent healthcare cloud provider, Oracle, is in a similar position as Google. The situation, however, could change in the coming years as relationships with key health IT vendors, such as EHR vendors, could alter this dynamic. For instance, Epic’s recent partnership with Google Cloud and Oracle’s acquisition of Cerner could impact the future choice of cloud service providers used for gen AI.

Foundation Models

Foundation models power gen AI. Examples of these models include large language models (LLMs), Multi-Modal Models (MMM), large vision models (LVM), and large audio models (LAudiMs).

It is also important to note that there are open-source and proprietary models. Open-source models are made freely available for possible modification and redistribution in the community, offering transparency and enabling collaboration. The code can be inspected, modified, and contributed to by anyone, which fosters a community-driven development environment. This can result in faster innovation and a diverse range of applications. Furthermore, open-source licensing allows software to be used freely without requiring users to pay fees or abide by strict usage restrictions. A fee-free gen AI model could be beneficial compared to a paid option.

In comparison, proprietary models are developed and maintained by private companies. These models often come with dedicated teams of experts who provide robust support and continuous improvement. Companies may invest heavily in research and development, resulting in cutting-edge technologies not immediately available in open-source alternatives.

Most FMs are ‘domain-agnostic’ – in other words, pre-trained on millions of generic data points, not just healthcare-related information. Examples of this type of FM are GPT-4 (OpenAI), Amazon Titan (AWS), PaLM 2 (Google), and Claude 2 (Anthropic), among others. The other type of FM is ‘domain-specific’. As the name implies, this FM is fine-tuned to a particular industry. This is a much less developed space than its domain-agnostic equivalent. However, examples are beginning to gain prominence, including Google’s Med-PaLM-2 and Hippocratic AI.

Healthcare providers are beginning to incorporate AI solutions into their services. However, as domain-specific models are still in the early development or preview stages, healthcare providers are using commercially available, proprietary, domain-agnostic models (mostly LLMs). Currently, the applications of these models are limited to a handful of use cases, which will be discussed in the following Insights.

Model Hubs / MLOps & LLMOps

Once the model is developed and ready to be used, a model hub allows models to be stored, shared, and managed. The Machine Learning Operations (MLOps) platform goes a step further and integrates the practices and tools used to streamline and automate the machine learning lifecycle (depicted below) and accelerate the process of building, training, and deploying models.

Figure 3: Machine Learning Modelling Cycle

For MLOps, the comparable products are Vertex AI from Google, AWS’s SageMaker JumpStart and Azure ML, all of which offer pre-trained ML models, ML frameworks, and other AI tools and resources for developers to utilise in model training and inference.

For gen AI, the products are not necessarily comparable in the same way, although there is some overlap in the offering. AWS Bedrock is a fully managed service that makes leading FMs, created by AI21 Labs, Anthropic, Cohere, Meta, Stability AI, and Amazon, available through an application programming interface (API) in addition to a broad set of capabilities that allow users to build and scale exclusively gen AI applications quickly.

Azure AI services is a comprehensive suite of out-of-the-box and customisable artificial intelligence tools, including LLMs, machine learning frameworks, and pre-trained models. It is a broader umbrella term that encompasses a wide range of services. Among these services is Azure OpenAI Service, a cloud-based platform that grants access to OpenAI’s powerful language models, such as GPT.

Vertex AI is similar to Bedrock and Azure OpenAI Service, providing a cloud-based platform for building and deploying AI applications. However, it doesn’t have a separate gen AI offering; instead, Vertex AI provides an ML platform with distinct gen AI capabilities integrated into it. The graphic below depicts how Google segments its Vertex offering.

Figure 4: Vertex AI Platform

Azure and Google have an extensive offering of open-sourced models on their websites, whereas Bedrock does not, although SageMaker JumpStart does. All platforms enable developers to adjust and optimise the models to meet their needs, and all offer scalability in their product mix.

The key difference is that AWS Bedrock offers a more extensive collection of models through its
partnerships. In contrast, Azure and Google leverage their proprietary and open-source models.
It’s essential to keep in mind that there are other options available. Hugging Face provides a model hub that is solely dedicated to open-source models.

This is a crucial aspect to consider for those contemplating the deployment of a gen AI solution. There is a high probability that open-source models will eventually catch up to their proprietary counterparts, making it difficult for developers of proprietary models to justify their premium pricing.


In terms of applications, all players established a gen AI market offering in the healthcare sector. Although some products are in the preview or beta testing stage, others have recently been released into general availability (GA). Microsoft has the highest number of applications in the pipeline, summarised in the table below.

Figure 5: Microsoft’s Gen AI Applications in Healthcare

AWS offers a range of healthcare products, namely HealthOmics, HealthLake, HealthImaging, and HealthScribe. Among these, HealthScribe is the latest addition and is currently in preview. It is the only gen AI-powered solution that AWS has designed for the healthcare industry so far. HealthScribe works by automatically creating clinical notes from conversations between patients and clinicians, thus reducing the workload on healthcare providers. It competes directly with DAX Copilot.

Lastly, Google has recently launched Vertex AI Search for healthcare, which employs medically tuned gen AI to facilitate users with searching through a wide array of data sources, including clinical notes, FHIR data, and electronic health records. This will be made possible by integrating with Google Cloud’s Healthcare API and Healthcare Data Engine, as well as with Google Health’s search and intelligent summarisation capabilities from Care Studio.

It’s essential to remember that big tech is constantly changing and introducing innovations. Therefore, if one organisation has a solution that another doesn’t, the latter will probably offer a comparable tool soon.

Now We’re Here

We have thoroughly examined the products and services the three major tech companies offer in generating artificial intelligence. Our analysis covers all aspects of the value chain, from the hardware infrastructure to the applications. The following graphic provides a comprehensive overview of the offerings from these three tech giants.

Figure 6: Big Tech Offerings in Gen AI Value Chain

It’s evident that all three organisations have a significant presence in every aspect of the gen AI value chain, with Azure being an exception in the hardware market. Big Tech will remain a crucial part of the ecosystem, providing all other aspects of product delivery, from hardware to end-user applications, even when open-source FM models catch up to their proprietary counterparts.

Throughout this series, which provides extract analysis from our Generative AI in Healthcare Market Intelligence Service, we will explore various applications of healthcare settings in greater detail. As demonstrated in the diagram below, we will classify and define the market based on application areas and use case clusters. We will identify use cases currently in use, under development or in the conceptualisation stage and provide insights on where we believe the market is headed. This will assist you in your strategic planning and help you identify key priorities.

Figure 7: Application Areas and Use Case Clusters in Health Care Network

Stay informed about the dynamic healthcare IT landscape with our AI-powered newsletter. Gain insights into the competitive field of generative artificial intelligence by subscribing to our newsletter.

Related Research 

Our new Generative AI Market Intelligence Service provides a rolling 12-month coverage of the fast-moving Generative AI market. It includes six market reports over this period examining:

  • Healthcare IT Vendor Opportunities and Strategies
  • Generative AI – Key Use Cases and Applications
  • Detailed Analysis of Generative AI Trends Observed and New Products Announced at Key Industry Events (e.g. HIMSS, RSNA)

About Vlad Kozynchenko

Vlad joined Signify Research in 2023 as Senior Market Analyst in the Digital Health team. He brings several years of experience in the consulting industry, having undertaken strategy, planning, and due diligence assignments for governments, operators, and service providers. Vlad holds an MSc degree with distinction in Business with Consulting from the University of Warwick. 

About the Digital Health Team 

Signify Research’s Digital Health team provides market intelligence and detailed insights on numerous digital health markets. Our areas of coverage include electronic medical records, telehealth & virtual care, remote patient monitoring, high-acuity clinical information systems, patient engagement IT, health information exchanges, Generative AI and integrated care & value-based care IT. Our reports provide a data-centric and global outlook of each market with granular country-level insights. Our research process blends primary data collected from in-depth interviews with healthcare professionals and technology vendors, to provide a balanced and objective view of the market.     

About Signify Research

Signify Research provides healthtech market intelligence powered by data that you can trust. We blend insights collected from in-depth interviews with technology vendors and healthcare professionals with sales data reported to us by leading vendors to provide a complete and balanced view of the market trends. Our coverage areas are Medical Imaging, Clinical Care, Digital Health, Diagnostic and Lifesciences and Healthcare IT.

Clients worldwide rely on direct access to our expert Analysts for their opinions on the latest market trends and developments. Our market analysis reports and subscriptions provide data-driven insights which business leaders use to guide strategic decisions. We also offer custom research services for clients who need information that can’t be obtained from our off-the-shelf research products or who require market intelligence tailored to their specific needs.

More Information

To find out more:
T: +44 (0) 1234 986111