Nvidia plans to launch RTX 60 GPUs codenamed Rubin in 2028

Nvidia’s next big leap for PC gamers is getting further away. The company’s GeForce RTX 60 series, built on an architecture codenamed Rubin, is now targeting a 2028 launch window, according to leaks from sources with a strong track record on GPU roadmaps.

That’s a meaningful shift. Earlier estimates had placed Rubin consumer GPUs somewhere in 2027, with some projections pointing to the first half of that year. Those timelines have since been walked back, and the 2028 target is now what well-regarded leakers Kopite7kimi and Kepler_L2 are pointing to.

Why the wait

The short answer is that Nvidia has a more lucrative problem to solve right now. TSMC, which fabricates Nvidia’s chips, is under enormous production pressure from AI accelerator demand. Nvidia’s data center business has become its dominant revenue engine, and when you have to choose where your fab capacity goes, AI chips win that argument on margin every time.

The Rubin architecture already exists in a data center form. Its consumer variant, the kind that goes into gaming PCs, hasn’t been formally announced. Nvidia is essentially asking gamers to wait while it services the market that generates far higher returns per wafer.

For context, Rubin’s data center specifications are genuinely staggering: 336 billion transistors and support for HBM4 memory up to 288 GB per GPU.

The practical consequence for gaming hardware is that the current RTX 50-series, built on Blackwell, gets a longer runway than originally expected.

Nvidia isn’t alone in the waiting room

AMD is in a similar position. The RDNA 5 lineup is also largely expected to arrive in 2028. One chip, referred to as AT2, may surface in 2027 as an exception, but the broader RDNA 5 rollout appears to be tracking on roughly the same delayed schedule.

That symmetry matters for how you read the competitive landscape. This isn’t a situation where Nvidia slips and AMD swoops in with a product advantage. Both companies are navigating the same supply chain reality, and both are feeling the gravitational pull of AI workloads on available fab capacity.

What the extended cycle actually means

Longer product cycles aren’t entirely bad news. When a GPU generation sticks around longer than planned, software tends to do more of the heavy lifting. Nvidia’s DLSS technology, which uses AI to reconstruct image quality at lower rendering costs, becomes more important when the underlying hardware isn’t refreshing on a predictable cadence.

The semiconductor investment angle here is straightforward in concept. Nvidia’s pivot toward AI production is already reflected in its financial results, where the data center segment has become the dominant revenue contributor. Investors watching GPU launch cadence as a signal for gaming revenue should be modeling a longer gap between Blackwell and whatever Rubin delivers for consumers.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.