Having spoken at the recent AI Expo event in London https://www.ai-expo.net/global/ attended by over 10,000 people there were some common themes amongst the attendees and speakers alike.
- For those already testing and deploying AI/ML projects there was some frustration with constraints in public cloud platforms on data transfer rates between CPU and GPU which was also constraining the model sizes that could be effectively utilised (whether image resolution or matrix size).
- Many companies that were looking to start a PoC or test model were looking for a cost effective platform where additional support could be provided for guidance and hand holding and where a final model could be deployed in a UK data centre rather than public cloud. Of course interfaces were needed into apps deployed in public clouds and external data feeds but the core application was preferred to be deployed in a UK environment.
- Retail (ecommerce and hospitality) as well as Healthcare seemed common industries who had already started or were ready to start AI/ML initiatives. Finance and Insurance seemed to already have PoCs or initiatives underway but were being frustrated by the constraints mentioned above.
- Established companies seemed to be looking to make more sense of their internal corporate data to facilitate more decision making based on fact rather than opinion while startups and innovators (by definition) were creating applications to enhance their focus industry (or disrupt depending on your point of view).
At L3C our AI/ML UK cloud platform is based on IBM Power9 technology and PowerAI which enables faster data transfer, larger model support and scalability without sacrificing performance thereby overcoming some of the frustrations the attendees were expressing and our discussions were enlightening and informative. To find out more about our AI/ML cloud platform and ease of migration see https://l3c.cloud/services/ai/ or contact us at firstname.lastname@example.org