AI’s Memory Boom Could Run the Self-Driving Car Revolution Off the Road

Artificial intelligence is creating winners and losers across the global economy. Investors have spent the last two years focused on the obvious beneficiaries — Nvidia (NASDAQ:NVDA | NVDA Price Prediction), Micron Technology (NASDAQ:MU), and the hyperscale data center operators spending hundreds of billions of dollars on AI infrastructure. 

Yet every boom creates unintended consequences. A recent Deutsche Bank report suggests AI’s insatiable demand for memory could create shortages that ripple far beyond Silicon Valley. One surprising casualty may be the autonomous vehicle industry, which relies on many of the same memory chips now being diverted into higher-margin AI systems.

AI Is Starting a Memory Arms Race

For years, semiconductors were the limiting factor in advanced computing. Today, memory is becoming just as important.

According to Deutsche Bank, demand for both high bandwidth memory (HBM) and conventional DRAM is expected to outpace supply for years as AI adoption accelerates. Every advanced AI accelerator requires large amounts of memory to process and store data. As a result, manufacturers such as Micron, Samsung, and SK hynix are increasingly allocating production capacity toward AI customers.

The challenge is that memory production cannot be expanded overnight. Building and equipping a semiconductor fabrication plant often requires investments exceeding $10 billion and several years of construction. As AI companies absorb more available supply, other industries could find themselves competing for what’s left.

Autonomous Vehicles Need Massive Amounts of Memory

The autonomous driving revolution was already facing obstacles ranging from regulation to consumer adoption. Memory shortages could add another roadblock.

According to Micron, future Level 4 autonomous vehicles may require more than 300 gigabytes of memory. That’s multiple times more memory than today’s vehicles use and far more than traditional automotive applications required just a decade ago.

A self-driving vehicle continuously processes data from cameras, radar systems, lidar sensors, mapping software, and onboard AI systems. Every one of those functions consumes memory. Meanwhile, AI data centers are becoming memory-hungry monsters.

A single advanced AI server can contain terabytes of memory, and major technology companies are expected to spend hundreds of billions of dollars annually on AI infrastructure over the next several years. Faced with that demand, memory manufacturers naturally prioritize customers willing to pay the highest prices. That leaves automakers in a difficult position.

Higher Costs Could Slow Autonomous Adoption

Surprisingly, the biggest threat may not be a lack of memory altogether. It may be the price. If automakers must pay more for advanced DRAM and HBM components, the cost of autonomous systems rises alongside it. Manufacturers then face two choices:

• Absorb the higher costs and accept lower margins.
• Pass those costs to consumers through higher vehicle prices.

Neither outcome is ideal. According to the Federal Reserve Bank of New York, the average new car loan balance already exceeds $40,000. Adding thousands of dollars in advanced computing hardware could make autonomous features harder to market to budget-conscious buyers.

That could slow adoption just as the technology is beginning to gain traction.

Key Takeaway

In short, AI’s memory boom may create an unexpected bottleneck for autonomous vehicles. Deutsche Bank’s research suggests memory demand could remain tight through the end of the decade as AI infrastructure absorbs a growing share of global production.

Granted, automakers have navigated semiconductor shortages before. Yet this challenge differs from the pandemic-era chip crunch because AI demand continues expanding rather than fading.

Ultimately, autonomous vehicles and AI data centers are competing for many of the same resources. If memory manufacturers continue directing capacity toward higher-margin AI customers, the road to self-driving cars may become longer and more expensive than the auto industry expected.