Gartner Says 40% AI Projects Will Fail, Memory Shortages Test Chipmakers

The article argues that successful AI deployments depend more on disciplined implementation—strong data pipelines, usage controls and clear governance—than on sheer spending on infrastructure. Chipmakers lacking robust project management frameworks risk underutilized hardware and stranded assets.

Research firm Gartner warns that more than 40% of all AI initiatives launched by 2027 will be terminated prematurely due to inadequate data quality, governance lapses and insufficient organizational alignment. This projection has heightened investor focus on execution metrics over capacity additions.

Ongoing global memory shortages and supply-chain bottlenecks are compounding execution challenges by delaying hardware availability and inflating component costs. Vendors with long-term memory contracts and diversified supplier networks are positioned to maintain production schedules and margins.

AMD’s recent multi-year memory supply agreements and emphasis on cross-functional governance could mitigate execution risks identified by Gartner. Its integrated approach to data management and project oversight may help capture a larger share of AI workloads as other vendors struggle with cancellations.