Hitachi Vantara turns to AI as storage supply crunch bites

Hitachi Vantara manufacturing plant floor
A glimpse at the floor of Hitachi Vantara's manufacturing plant in Oklahoma. (Diana Goovaerts/Fierce Network)
  • Hitachi Vantara is using AI to navigate a deepening storage supply crunch
  • Executives say component shortages could persist for 12+ months, forcing vendors to prioritize which customer deals get scarce supplies first
  • The company’s AI-driven forecasting and inventory tools are helping it allocate supply, manage lead times and cut inventory

NORMAN, OKLAHOMA – Hyperscale capex budgets aren’t the only ones feeling the AI-driven supply chain crunch. Vendors like Hitachi Vantara are feeling the heat, too, as spiking demand tightens availability of key components in the products they sell.

During a tour of Hitachi Vantara’s Oklahoma manufacturing plant, executives told Fierce Network that component constraints are extending lead times and forcing vendors to make sharper calls about which customer deals get scarce supplies first. But Hitachi is betting AI can help solve that problem by improving demand forecasting, prioritizing supply allocation and eventually reshaping how its own manufacturing operations run.

Hitachi Vantara supplies the IT stack – hardware, software and more – for the storage systems that are foundational for AI. That means memory like DRAM and NAND as well as solid state drives (SSDs) are core parts of its offering. Those components are in extremely short supply at the moment.

Using AI to navigate constraints

Yuichi Fukuda, Hitachi Vantara’s chief operations and global supply chain officer, characterized the supply situation as “very tight”. He noted that lead times for SSDs have doubled from three months to six months over the past year. 

With demand rising as component supply contracts, Fukuda said it’s more important than ever for the company to stock the right amount of inventory. Stocking too much can be costly, while stocking too little can mean painful waits for customers. 

This is where the company is leaning on AI to help. Hitachi Vantara has built a deal prediction tool that scans data inputs in Salesforce daily to predict what deals will close. Those predictions – which are around 80% accurate – help inform inventory decisions. The company has also spun up what it calls the “Inventory Control Tower,” a tool it uses to unify inventory visibility, plan resource allocation with AI and stay ahead of supply disruptions. This has allowed it to slash its inventory by 50% over two years. 

Additionally, with a lot of customer deals on its hands and tight supply “we need to think about which deal we should prioritize,” Fukuda said. “We use AI to make prioritization of the order deals, and we allocate the supply to the higher priority deals, customers. We started that kind of thing to optimize our delivery to the customer.”

Tena Coppedge, Hitachi Vantara’s VP of Global Supply Chain Manufacturing, said these tools also help manage customer expectations.

“We need to be able to give our customers realistic due dates and realistic expectations,” she said. “And I feel like if we can do that and we can do that consistently, then even though it may be an extended lead time, they're kind of prepared for that in the situation that we're in today.”

Supply and demand for networking components

These kinds of workarounds matter because the pressure isn’t short-term in nature. 

Coppedge said parts and component shortages are expected to persist over the next 12 months. Fukuda said DRAM constraints could last even longer. And demand isn’t necessarily falling off, even despite the kinks in the supply chain and rising prices. 

Hitachi’s customer base is primarily comprised of enterprise customers across verticals like telecom, finance, healthcare and government rather than hyperscalers. Simon Ninan, Hitachi Vantara’s SVP of business strategy, said while some smaller customers are pulling back because they can’t afford the higher prices, large enterprises are actually buying now to lock in costs before they go even higher. 

Hyperscalers are effectively “locking up” demand with massive growth plans, while fabs don’t have enough near-term capacity to fully meet that appetite, he said. More manufacturing capability is coming online, he noted, but it likely won’t meaningfully ease the market in the next 12 months; relief may be more of an 18- to 24-month story. 

That leaves enterprise infrastructure vendors and customers navigating a market where component supply remains tight, prices are under pressure and the pace of hyperscaler AI capex could determine how quickly — or slowly — the broader market normalizes.

Read more about memory shortages on Fierce Network

Mobile ecosystem is facing major memory problems

Memory shortage puts the squeeze on telco modernization efforts

Broadband groups warn White House memory prices could upend supply chain

Virtualized memory could be the key to better AI performance