AI in Supply Chains
Fragile by Design: What AI Can and Can't Fix in Supply Chains
Asrita Yelisetti
AI has slowly become the answer to almost every business problem. Is customer service too slow? Implement a chatbot. Need to optimise costs? Replace functions with AI. Supply chain processes are no exception. In fact, it’s become one of the biggest frontiers for artificial intelligence adoption. However, AI is a powerful tool that’s being used to fix a system that was deliberately made to be fragile in the first place. And understanding this difference is essential for the next generation of business leaders.
Companies are using machine learning models to predict demand with far more accuracy than traditional forecasting ever could. According to McKinsey, AI-driven forecasting reduces demand errors by 20-50% compared to traditional methods. This directly leads to a decrease in wasted inventory and fewer stockouts. Retailers like Walmart are already using AI to analyse local shopping patterns and weather data to adjust their inventory before a customer even realises that they need something.
The market is growing at an incredible speed, as it’s projected to go from $13.9 billion in 2025 to $50.41 billion by 2032. That’s a 20% annual growth rate. McKinsey’s data on AI distribution operations shows reductions of 20-30% in inventory costs, 5-20% in logistics costs, and 5-15% in procurement spend for companies that implement it well. These reductions could mean saving hundreds of millions of dollars a year.
Beyond forecasting, AI is also being used for route optimisation, warehouse automation, and supplier risk monitoring. AI in supply chains is real and impactful.
When the COVID-19 pandemic hit, global supply chains essentially collapsed, exposing a decades-long design flaw. The 2021-2023 global supply chain crisis was not solely because of the pandemic. It was the result of companies spending decades optimising for efficiency at the expense of resilience. The “just-in-time” manufacturing philosophy is dependent on keeping minimal inventory and relying on perfectly timed deliveries, however, there was no buffer when factories got shut down, ports got congested, and consumer demand spiked literally overnight.
The global chip shortage is a clear example. Automakers and other industries had slashed chip orders when factories shut down at the beginning of the pandemic, but when the demand rebound was faster than expected, chipmakers couldn’t keep up. Multiple companies, including Ford, Volkswagen, and Tesla, were forced to cut production for months due to a weak supply chain.
AI is solving problems that were created by bad strategy, not bad technology, and while AI can improve a system efficiently, it can’t entirely make a fundamentally fragile system resilient. Biased or incomplete data causes flawed AI predictions, and to this day, multiple companies still struggle with fragmented data sources across departments. Only 23% of supply chain organisations have a formal AI strategy, yet 85% increased AI investment last year. This gap between spending money on AI and actually knowing how to use it reveals an underlying issue with strategy that many companies face.
There’s also another cost to consider when looking at AI’s ability to increase efficiency: workers. Amazon’s automation has already reduced warehouse staff needs by 20-25%, and with self-driving trucks in the near future, hundreds of thousands of long-haul trucking jobs could essentially disappear. Many of these people already work difficult and physically demanding jobs, yet lack access to retraining programs that would prepare them for AI-adjacent roles. While AI will create new jobs, as the World Economic Forum projects that AI will displace 92 million jobs globally by 2030, but create 170 million new ones. Roles that barely existed years ago are currently being hired for. But these new jobs disproportionately go to people with tech skills and higher education - not necessarily displaced workers. The 170 million new jobs that AI will create don’t show us who benefits and who will be left behind. This poses the question: whose responsibility is it to ensure that these workers aren’t left behind? While there’s definitely somewhat of a responsibility on each individual, given that they have the resources to keep up with the game, there’s also a responsibility on these multi-million dollar corporations to take care of their employees.
AI is not going to save supply chains by itself. It reflects man-made problems in the systems that were set up and can be used as a solution for these problems. The lesson to be learned from the last few years isn’t that we need smarter algorithms, it’s that we need smarter design. Supply chains that are built for resilience will benefit enormously from AI tools layered on top. Fragile supply chains are still prone to collapsing, even with AI integration. Optimisation isn’t the same as strength. AI is a tool, but design is still on us.
Sources:
https://www.marketsandmarkets.com/Market-Reports/ai-in-supply-chain-market-114588383.html
https://en.wikipedia.org/wiki/2021%E2%80%932023_global_supply_chain_crisis
https://expertnetworkcalls.com/95/causes-of-the-global-semiconductor-shortage-and-its-impact
https://www.scoperecruiting.com/blog/supply-chain-roles-replaced-by-ai-2026
