Rivian Pays Premium Prices for Memory as AI Data Centers Squeeze Supply

Rivian says it has already paid premium pricing to secure memory capacity because AI data centers are swallowing supply the auto industry once treated as a commodity, and the EV maker may face further delays securing parts. The disclosure sits in Rivian's third-quarter 10-Q, filed October 29, and CEO RJ Scaringe has now put numbers of a different kind behind it: hours of engineering time.

"Three years ago, we were not spending a lot of time thinking about memory. It was a commodity," Scaringe said in a conversation with venture firm Kleiner Perkins, reported by InsideEVs on November 11. "And now an enormous amount of time goes into planning for memory, optimizing it at the board level, how we're using it, and then really thinking strategically about how we source it."

What did Rivian tell investors?

The 10-Q is blunt. AI development "has created unprecedented demand for compute infrastructure, memory and related hardware and power, and vendors may prioritize other AI developer customers and contracts," the filing states. "As a result, we have had to, and may continue to have to, pay premium pricing to secure capacity and may experience delays in securing such capacity."

Rivian has not quantified the added cost. Nor has it tied any specific production change to the squeeze. But the filing language — "have had to" — confirms the premiums are already being paid, not merely forecast.

Which parts beyond memory are short?

Memory is not the only pinch point. Vidya Rajagopalan, Rivian's hardware chief, said the data center boom has created shortages in far more mundane components.

• E-glass for printed circuit boards: "With the data center boom, a lot of the e-glass for [printed circuit boards] is really hard to find. And so you could have a component shortage in one space that totally throws you off," Rajagopalan said.
• Multilayer ceramic capacitors (MLCCs): These basic circuit-board parts, long invisible to vehicle planners, are now on Rivian's radar. "MLCC capacitors, you never thought about them. You're thinking about them now, right?" she said.

The pattern is consistent: AI server demand pulls glass weave, capacitors and DRAM toward hyperscale buyers, leaving automotive programs — historically low-margin customers for commodity parts — at the back of the allocation queue.

What does this mean for the R2?

On Rivian's latest earnings call, Scaringe said the company is hunting for offsets to higher memory and other commodity costs so the R2 can hold its advertised $45,000 base price, InsideEVs reported. That target matters commercially: the R2 is Rivian's volume play, and memory inflation lands squarely on its bill of materials.

The R2 is also due to receive Rivian's in-house RAP1 autonomy processor next year, a part whose production ramp adds another layer of silicon planning to the same stressed supply chain.

The same Kleiner Perkins conversation produced another data point: Scaringe said Mind Robotics machines will first take on about 20% of tasks in Rivian's plant, per MLQ.ai's coverage of the remarks.

Are other carmakers feeling it?

Yes, and they're moving on the same playbook. GM and Ford both cited higher memory costs on recent earnings calls and have signed long-term supply deals with Micron, locking in allocation the way PC makers once did during past DRAM cycles. China's Li Auto went further, warning of "significant supply tightness and price volatility" in both DRAM and NAND.

The margin side of the ledger tells the rest of the story. Memory makers' margins have swelled as AI servers absorb supply, MLQ.ai reported, and the cost pressure has reached consumer graphics cards as well — evidence that the AI buildout is repricing silicon across every downstream market, not just automotive.

For Rivian, the near-term question is whether strategic sourcing and board-level optimization can outrun memory pricing through the R2's launch window. The company has absorbed premiums so far without announcing production changes; whether that holds through next year's RAP1 introduction and R2 ramp will test whether a mid-scale automaker can buy its way out of a cycle built for hyperscalers.