The Moore Threads Lushan GPU could become the highest-capacity VRAM consumer graphics card ever released, according to a product manager at the Chinese chipmaker. The claim appeared on Weibo on November 18, but Moore Threads has not confirmed an actual memory figure, leaving the headline number a mystery for now.
Key takeaways
- ✓A Moore Threads product manager claims the Lushan GPU could set a new consumer VRAM record
- ✓The exact VRAM capacity has not been disclosed, despite the public claim
- ✓Moore Threads previously promised four times the memory of the 16GB MTT S80, implying a possible 64GB figure
- ✓The GeForce RTX 5090 currently tops out consumer VRAM at 32GB
- ✓Lushan is expected to be detailed at the MUSA Developer Conference in Beijing on December 29
Moore Threads followed up the social media tease with its first official poster for MDC 2026, the MUSA Developer Conference scheduled for December 29 in Beijing. The artwork shows two mountain peaks, one rendered as a 3D wireframe mesh and the other styled as a neural network, with the card itself largely obscured behind them. It is a deliberately vague reveal, but it confirms the company intends to use the conference to properly unveil Lushan.
Table of Contents
What the Moore Threads Lushan GPU might offer
Moore Threads first floated the concept behind Lushan at the inaugural MUSA Developer Conference a year earlier, alongside a new architecture called Huagang. At that event, the company made some striking performance claims, including 15 times higher performance in AAA gaming titles compared with its previous generation, 50 times faster ray tracing, and 64 times more compute throughput. Those are enormous jumps for any single architectural leap, and they set expectations high for whatever Moore Threads eventually brings to market.
The current flagship in the lineup, the MTT S80, ships with 16GB of VRAM. Moore Threads has previously said its next generation would carry four times the memory capacity of that card, which works out to 64GB if the company follows through. That figure is speculation rather than confirmed fact, since no official VRAM number has been published alongside the record claim. Still, if Moore Threads does land anywhere near that mark, it would comfortably beat every mainstream consumer card on the market today.

How the VRAM race stacks up so far
The benchmark Moore Threads appears to be targeting is Nvidia’s GeForce RTX 5090, which currently holds 32GB of VRAM, the largest capacity on any widely available consumer graphics card. Whether the RTX 5090 genuinely counts as a mainstream consumer product is itself a point of debate, given its premium positioning, but it remains the figure most reviewers use as the ceiling.
| Graphics card | Maker | VRAM | Status |
|---|---|---|---|
| MTT S80 | Moore Threads | 16GB | Current generation |
| GeForce RTX 5090 | Nvidia | 32GB | Current record holder |
| Lushan (speculated) | Moore Threads | Up to 64GB (unconfirmed) | Expected reveal December 29 |
Moore Threads is a relatively young entrant compared with Nvidia, AMD, and Intel, and it has built its reputation largely on the domestic Chinese market, where it has positioned itself as a homegrown alternative amid ongoing export restrictions on foreign silicon. A record-breaking VRAM claim, even an unconfirmed one, is a useful way to grab attention ahead of a major local showcase like MDC 2026.
Why so much memory matters, and who actually needs it
Large VRAM pools matter for more than just bragging rights. Modern games at high resolutions, with ray tracing and uncompressed texture packs enabled, can chew through memory quickly, and a GPU VRAM shortage already squeezing the market has made every extra gigabyte feel more valuable than it used to. However, the gap between what gamers actually require and what a 64GB card would offer is enormous, which raises an obvious question about who Lushan is really built for.
Workstation users running large creative or engineering workloads, along with hobbyists experimenting with local large language models, are often the ones pushing memory capacity to its limits rather than gamers. A card with that much VRAM would suit those crowds just as well, if not better, than anyone chasing frame rates. That tension between gaming and compute use cases is becoming more common across the industry, as seen in AMD’s own roadmap leaks for its Zen 6 Medusa GPU and Intel’s reported 480GB AI GPU, both of which blur the line between consumer and data centre hardware.

Until Moore Threads confirms a real specification sheet, the Lushan VRAM figure remains an educated guess built from a company executive’s own prior statements. The MUSA Developer Conference on December 29 should finally settle the matter, alongside whatever pricing and availability details follow. For now, the teaser poster and the Weibo post are enough to put Lushan on the radar of anyone watching how Moore Threads plans to challenge the established graphics card makers, much as Intel’s own Crescent Island GPU has done on a different front.
Frequently asked questions
How much VRAM will the Moore Threads Lushan GPU actually have?
Moore Threads has not confirmed a figure. A product manager claimed it would set a consumer VRAM record, and past statements about quadrupling the MTT S80’s 16GB suggest a possible 64GB, but this remains unverified speculation until the company publishes official specifications.
When will Moore Threads reveal the Lushan GPU?
Moore Threads is expected to detail Lushan at the MUSA Developer Conference, known as MDC 2026, which is scheduled for December 29 in Beijing.
Is the Moore Threads Lushan GPU aimed at gamers?
That isn’t entirely clear. While it is being framed as a consumer card, a VRAM pool that large would likely also appeal to workstation users and people running local AI models, who tend to need more memory than typical gaming workloads require.
What GPU currently holds the consumer VRAM record?
Nvidia’s GeForce RTX 5090 currently has the largest VRAM capacity among widely available consumer graphics cards, at 32GB.



