Through partnering with a prominent hyperscaler, OpenAI, the research laboratory behind the popular ChatGPT, intends to increase access to its AI models. This approach applies the principle of AI democratisation which is being adopted by participants throughout the space. Adherents advocate for increasing the accessibility and affordability of advanced AI technology for a broad range of users, including individuals and small businesses. The main players in the hyperscaler space have all publicly stated their commitment to the democratisation of AI, supported by taking steps to make their AI models and tools more readily accessible.

Whilst democratisation does not mean access is always without cost, the increased availability of AI toolkits are slated to enable a new generation of startups to leverage these AI models and tools to launch innovative businesses and create new revenue streams. Commercial partnerships between leading AI research laboratories, such as OpenAI and DeepMind, with the leading hyperscalers will make it easier for businesses to access the most advanced and effective AI technology.

The hyperscalers have three primary tools by which they allow access to their AI toolkits: their cloud-based services; open source licensing and APIs.

One of the most effective mechanisms being utilised by the hyperscalers is the making available of their AI models and tools through their cloud-based services. For example, Google’s TensorFlow machine learning library is made available through its cloud offering. This enables data scientists and developers to easily train, deploy, and manage their AI models through their existing cloud service arrangements without having to develop the AI models themselves or to invest in the hardware that provides the compute power required to run these tools.

A second approach is to make AI toolkits publicly available under an open source licence in a similar manner to the traditional approach to open source software. For example, hyperscalers have previously released the source code for their distributed deep learning toolkits under the ‘MIT Licence’, a popular open source licence.

A third approach is through the use of APIs. OpenAI provides access to its AI models via a general purpose “text in, text out” interface and intends to make use of the OpenAI API in connection with its hyperscaler partnership to allow more users to access the services. Of particular relevance to entrepreneurs looking for synergies with their existing products and the latest in AI technology, it is possible for users to obtain permission to integrate the OpenAI API into their existing products as well as for use with new products. It does not take much to imagine how a tool like ChatGPT could be integrated into legacy applications, such as email or word processors.

We expect startups with access to new or previously untapped data sets will utilise these tools to build creative new applications to, for example, gain valuable insights for their clients; ushering in an exciting new era for innovation across multiple industries. Whilst the access to these powerful technologies is creating a pressure for startups to find the most effective use case, those businesses engaging in this space should remain cognisant of the legal and operational risks inherent in this business model, such as understanding the contractual terms with their hyperscalers, vendor lock-in, regulatory compliance issues and how each of these issues will influence downstream interactions with their customers. Further analysis of the points to consider when leveraging generative AI in your business can be found here.