Businesses seeking growth through AI-driven initiatives are encountering challenges, as experimental efforts frequently fail to evolve into fully implemented projects.
AI, especially the latest generative AI models, are a transformational technology that could radically change the way we work. But these plans must be executed with a robust and comprehensive strategy to succeed.
According to Foundry’s 2025 AI Priorities Study, over half of businesses expect to increase their AI spending in 2025.[1] And almost half of organisations (49%) now have dedicated AI budgets, significantly higher than just two years ago.[2]
Yet industry analysts IDC says the “AI scramble” is unsustainable.
Businesses have found it hard to move from proof of concept to production and have not seen all the anticipated benefits of AI, according to industry analysts IDC.
Rick Villars, IDC’s Group Vice President, Worldwide Research, says: “Through April 2024, a typical enterprise identified hundreds of GenAI use cases. They launched dozens of proofs of concept (POCs), but they’d only put six into production. [3]
By January 2025, when IDC checked back in, companies reported that they’d only managed to get 44% of their GenAI projects into full production. This GenAI scramble is not sustainable for enterprises, or for technology providers who care about converting POCs into sustainable, long-term business.”[4]
Strategic alignment
These issues are frequently caused by insufficient strategic alignment between AI projects and the broader objectives of the enterprise, such as digitisation and modernisation.
Simple AI projects can deliver success in areas such as better customer service or quicker code development in IT. But AI will not transform enterprise operations unless it is part of both the organisation’s technology and business strategies.
Successful organisations see AI as part of a longer-term transformation of the business, with AI extending its reach as the technology develops and matures.
Hasmukh Ranjan is CIO and senior vice president at AMD, where he manages a growing number of AI projects. “The way we look at AI is in four stages of development: assist action, automate and autonomous,” he says.
“The most successful AI solutions today are ‘assist’, where co-pilots, chatbots, and agents are helping the user become more effective by quickly analysing data and ensuring they have the right information at their fingertips immediately.”
Ranjan adds some projects are now moving to the “action” stage, where AI agents are beginning to “take actions for different workloads and workflows.”
Layers of intelligence
Nevertheless, to fully leverage the advantages of artificial intelligence, organisations must adopt a long-term perspective.
Successful AI strategies incorporate multiple technology layers, from large language models (LLMs) to smaller, more focused AI tools running on personal devices.
“If you have a multi-year view where you can build your solutions and have a strategy which takes you from ‘assist’ all the way to autonomous systems, there is great value to be gained in every enterprise,” says Ranjan.
AI projects based on LLMs need large-scale GPU deployments in suitable data centres – ideally with liquid cooling, fast networking and extensive storage.
The latest generation of AI PCs, including those based on AMD Ryzen™ PRO processors, provide capabilities previously limited to cloud deployments.
These PCs offer assistance tools such as document classification or translation, or advanced data analytics, without the need to store sensitive data in the cloud. And the cost of that computing power continues to fall, says Ranjan, allowing organisations to consider bringing AI in house at all levels.
“A comprehensive AI strategy involves not only discussions at the application layer, but the entire stack,” he says.
Find out more about AMD’s AI solutions.
[1] Foundry AI Priorities Survey, 2025, Foundry
[2] Ibid
[3] Worldwide Monthly Technology Investment Monitor: January 2025: AI Adoption Priorities and Agentic AI Plans for Business Processes, IDC #US52334925
[4] Worldwide Monthly Technology Investment Monitor: January 2025: AI Adoption Priorities and Agentic AI Plans for Business Processes, IDC #US52334925