Nvidia and SK hynix announced a multiyear technology partnership on June 7 to co-develop the next generation of memory for the global “AI factory” buildout — a deal that binds the world’s most valuable chipmaker to its most important memory supplier at the exact moment ordinary buyers are paying more for the RAM inside their computers.
According to Nvidia’s investor announcement and a parallel statement from SK hynix, the two companies will jointly engineer memory for Nvidia’s Vera Rubin AI supercomputers, its Vera CPUs, RTX Spark-powered PCs and the Jetson Thor robotics platform. The agreement spans Nvidia’s entire roadmap — from the data-center systems that train frontier models down to the desktop and robot chips headed for consumers.
What Nvidia and SK hynix Actually Agreed To
The two companies framed the deal around three markets Nvidia says it is building: AI infrastructure, “personal AI” and “physical AI.” SK hynix said it will supply memory for all three — high-bandwidth memory (HBM) for Vera Rubin, memory for the RTX Spark desktop AI PCs Nvidia showed at Computex 2026, and memory for the Jetson Thor modules that run humanoid and industrial robots.
The pact also reaches into how the chips get made. The companies said they will use Nvidia’s CUDA-X libraries and PhysicsNeMo to speed up semiconductor simulation, and apply Nvidia Omniverse, OpenUSD and cuOpt to move SK hynix toward “fully autonomous” fabrication plants built on factory digital twins. The stated goal: compress the multi-year development cycle of advanced memory so supply keeps pace with Nvidia’s hardware roadmap.
Why This Reaches Your Desktop
The bottleneck in the AI boom is no longer just GPUs — it is memory. HBM, the stacked memory bolted next to AI accelerators, eats far more wafer capacity per gigabyte than the conventional DRAM that goes into phones, laptops and consoles. When memory makers pour capacity into HBM for AI customers, less is left for everyone else.
That is why a deal aligning SK hynix’s next-generation memory to Nvidia’s AI roadmap matters beyond the data center. SK hynix is already a confirmed HBM4 supplier for Vera Rubin, which entered full production earlier this month alongside Samsung and Micron. The June 7 agreement goes further, formalizing co-development and supply alignment across Nvidia’s whole stack rather than a single platform.
Are RAM Prices Going Up Because of This?
Not because of this deal alone — but it reinforces the trend. Memory prices have been one of the most consistent consumer-tech stories of 2026. AMD chief executive Lisa Su has identified high-bandwidth memory as the next constraint in the AI chip supply chain, DDR5 prices for PC builders have climbed sharply through the year, and analysts have warned of a steep memory cost surge across the cycle.
A multiyear pact that explicitly prioritizes AI-factory memory is, for a consumer, double-edged. More predictable supply and autonomous fabs could eventually ease shortages. But locking a top supplier’s next-generation capacity to AI customers first reinforces the same dynamic that has been raising the price of the RAM in your devices.
Why Nvidia Wanted a Tighter Memory Partner
Memory has become Nvidia’s strategic chokepoint: each new accelerator demands more HBM, faster, and the company cannot ship systems it cannot feed. Binding SK hynix to its roadmap — and helping automate the supplier’s fabs — secures that supply years ahead.
The “physical AI” and “personal AI” language also signals where Nvidia expects growth. Jetson Thor is its pitch for the coming wave of humanoid robots; RTX Spark is its attempt to put data-center-class compute on a desktop. Both need purpose-built memory, which the co-development arrangement is meant to deliver.
What You Can Do About It
If you are planning to build or buy a PC, upgrade a laptop, or grab a console in 2026, the practical read is unchanged: memory-driven price pressure is unlikely to vanish soon, and buying while prices are merely high — rather than waiting for them to climb further — remains the conventional hedge. The same partnership investing in autonomous fabs could expand capacity over the next few years, but any relief is a multi-year story, not a this-quarter one.
Neither company disclosed the financial terms, the duration, or specific volume commitments. What is clear is that the two companies most responsible for the hardware behind the AI boom have tied their futures together — and that the memory inside your next device is increasingly being designed with AI factories in mind first.
Frequently Asked Questions
What did Nvidia and SK hynix announce? On June 7, 2026, the two companies announced a multiyear technology partnership to co-develop next-generation memory for AI “factories,” spanning Nvidia’s Vera Rubin supercomputers, Vera CPUs, RTX Spark PCs and Jetson Thor robotics platform, plus AI-driven chip design and autonomous fab manufacturing.
Will this make RAM more expensive? The deal itself does not set consumer prices, but it reinforces the existing trend of AI demand prioritizing HBM and squeezing standard DRAM supply — the dynamic that has been pushing DDR5 and other RAM prices up through 2026.
Is SK hynix already supplying Nvidia? Yes. SK hynix is a confirmed HBM4 supplier for Nvidia’s Vera Rubin platform, alongside Samsung and Micron. The new agreement broadens the relationship into co-development across Nvidia’s full product range.
What are RTX Spark and Jetson Thor? RTX Spark is Nvidia’s desktop AI PC platform; Jetson Thor is its computing module for humanoid and industrial robots. Both are covered by the new memory co-development deal.