Nvidia reintroduces the RTX 3060 with 12 GB of video memory

Technical specifications remain unchanged

The old card thus enters into direct competition with significantly younger models. In games such as Assassin's Creed Shadows, Clair Obscur: Expedition 33, or Battlefield 6, the current RTX 5060 model is better positioned, achieving five to 20 percent more fps in almost all scenes shown. However, the RTX 5060 only has 8 GB of video memory, which can lead to stuttering in some games. „Dabei benötigt diese Generation offenbar weniger Leistung", the trade publication writes regarding the power consumption of the new generation.

In addition to Nvidia cards, buyers in this price range will also encounter AMD's RX 9060 and RX 9060 XT with 8 GB of VRAM each, as well as Intel's Arc B580, which according to retailer information is often offered for less than 300 euros. The Arc B580 has 12 GB of VRAM, is, according to the report's assessment, more powerful than the RTX 3060, and supports modern upsampling as well as frame generation with XeSS 3. AMD supports FSR upscaling in version 4.1 according to its own statements; „so berichtet Nvidia, dürfte AMD auch eifrig an einer Frame-Generation-Lösung arbeiten", it states in this regard.

Competition in the mid-range

A key disadvantage of the RTX 3060 is the lack of support for frame generation — that is, AI-generated intermediate frames for a smoother gaming experience. Since the Ampere architecture does not offer all AI functions for modern game upscaling, the reissue cannot keep up with the current RTX 5000 series in terms of frame rate capture. Manufacturer Asus lists an affordable dual-fan version „Dual" of the GeForce RTX 3060 starting at around 335 euros.

According to the report, the RTX 3060 is manufactured using Samsung's 8-nanometer process, which, according to the current state of affairs, shows no capacity bottlenecks. The freed-up capacities in modern production lines are expected to be used for the production of further data center chips. Blackwell graphics chips are reportedly manufactured by TSMC using the 4NP process.

Background: AI boom and memory prices

The background to this move is the ongoing AI boom: since late summer 2025, prices for memory chips have been rising sharply due to massive demand from AI manufacturers and data center operators. Dedicated graphics cards were temporarily difficult to obtain since the crypto hype during the COVID-19 pandemic; newly announced hardware was sometimes sold very quickly in so-called „paper launches" according to the report. A brief recovery in supply lasted only about two years before the AI boom began.

Nvidia is described in the current report as the world's most valuable company with a market capitalization of 4.6 trillion US dollars. At Computex, the company also announced a new chip and server generation called „Vera Rubin", which is intended for largely automated, scalable „AI factories". Whether the RTX 3060 reissue is part of this strategy, the report leaves open.

Outlook on availability and strategy

Observers interpret the reintroduction as an indication that Nvidia wants to secure market share in the mid-range in the short term without tying up its own manufacturing capacity in favor of affordable gaming GPUs. According to the report, the free production capacities are to flow into数据中心einsatz. Nvidia is thus positioning itself between price-sensitive gamers and high-margin AI infrastructure.

For those interested in purchasing, it remains unclear for the time being how long the allocations will last. Since the cards come from new supply volumes, availability is considered secure in the coming weeks; in the medium term, however, the model is likely to be discontinued again once supply and demand align.