Nvidia's next gaming GPU may not ship until 2028, leaker claims

Kopite7Kimi, a leaker with a historically strong record on Nvidia hardware, posted Sunday that Nvidia's next consumer graphics cards, the GeForce RTX 60 series, will not arrive until 2028. If he is right, the cards on store shelves today will be running on a chip-manufacturing process roughly six years older than the one Nvidia's data-center customers are buying. That is the structural read the date implies.

Kopite7Kimi's specific claim: Nvidia's GR20x chip, the silicon expected to anchor the RTX 60 family, is now a "product from 2028," pushing his prior late-2027 expectation by another year. "That means they have postponed again," he wrote on X.

Nvidia has not officially announced the consumer Rubin lineup, product names, or launch date. Tom's Hardware said it reached out to Nvidia for comment and had not received an on-record response. The outlet reported that Kopite7Kimi's track record is strong but that the post is labeled commentary, not Nvidia disclosure. VideoCardz separately framed the update as a second leaker confirming the 2028 slip; the second leaker's identity and weight are not fully verified in the source available to Type0, so this is mosaic-level corroboration rather than independent confirmation.

Three mechanisms in the report hold up better than a single date does.

The current GeForce RTX 50 series sits on TSMC's 4N process. Nvidia's Rubin accelerators use the newer 3N and 3NP. If the 2028 date holds, the cards on store shelves today will be running on a process node roughly six years older than the one Nvidia's data-center customers are buying. The structural gap between consumer and data-center nodes is what Tom's Hardware is reading into the slip.

Memory is the second mechanism. The report cites a specific constraint: 3GB GDDR7 VRAM chips, the same memory type already used in the RTX 5090 and in the RTX Pro 5500. Supply of those 3GB chips is being routed to Nvidia's AI and data-center products, where the margin on a 3GB GDDR7 die is dramatically higher than on a consumer GeForce card. A consumer GeForce launch cannot easily compete with that allocation right now.

The economic backdrop sets the third. Nvidia's own framing of Rubin is that it is "the fastest ramp in company history." The forward-looking expectation, as restated by Tom's Hardware, is that the company will make $20 billion from its Vera Rubin data-center line in Q3 2027. That number is Nvidia's own guidance, not actuals, but it sets the gravitational pull: every engineering hour, every 3GB GDDR7 die, and every 3N/3NP wafer not bound for data-center is a wafer not bound for the fastest ramp in the company's history.

A related RTX 50 Super Blackwell refresh the same leaker had previously described as ready has reportedly been sidelined. Three slips across the same product family, with a refresh reportedly ready but unpushed, point to resource allocation rather than a calendar accident. That read is consistent with the trade-outlet framing, including Tom's Hardware's "gaming takes a back seat" line, but it does not require assuming Nvidia has abandoned gamers as a stated strategy. The source supports a delay, not a departure.

For a reader planning a 2026 or 2027 PC build or hold-out, the practical read is structural: anyone waiting for a big generational leap in consumer graphics is looking at a longer wait and a longer stretch on current hardware. RottenWifi's restatement makes the same point with a more explicit caveat, calling a 2028 debut a "reported possibility, not an official NVIDIA roadmap."

The falsifier is short and worth naming. A single leaker post, a strong but not perfect historical record, no Nvidia confirmation, and a second-leaker corroboration that is only partially verified. If Nvidia breaks its silence with a date, or if a second well-sourced leaker walks the 2028 number back, the structural read still survives. The wafer economics and the memory allocation are the story, not the calendar.