NVIDIA Partners with Samsung to Revive RTX 3060 Production Amid GPU Shortage

In the annals of consumer technology, it is exceedingly rare for a manufacturer to pull a half-decade-old product off the shelf, dust it off, and send it back to the factory floor. Yet that is precisely what NVIDIA appears to be doing. According to reports originating from the Korean financial outlet Hankyung and corroborated by multiple supply chain sources, Samsung Electronics' foundry division is preparing to resume production of NVIDIA's GeForce RTX 3060 graphics card — a product that first shipped in February 2021 — with new units expected to reach board partners between March 10 and March 20, 2026 [1][2].

The decision is not nostalgia. It is triage. And understanding why NVIDIA is reaching back to the Ampere generation requires untangling a web of interconnected crises — from AI-fueled memory shortages to U.S. export controls — that have conspired to make affordable PC gaming hardware scarcer than at any point since the cryptocurrency mining boom of 2021.

The GPU Famine of 2026

The current shortage is unlike its predecessors. During the crypto boom, demand was the culprit: miners bought every card they could find. This time, the crisis is rooted in supply — specifically, in the raw materials needed to build modern GPUs.

At the center of the problem is memory. AI datacenters have consumed staggering quantities of high-bandwidth memory (HBM) and advanced DRAM, driving prices skyward. TrendForce projects that AI-equivalent consumption will account for nearly 20% of total global DRAM output in 2026, with HBM consuming 4x the wafer capacity of standard DRAM per gigabyte and GDDR7 requiring 1.7x [3]. The three major memory manufacturers — Samsung, SK Hynix, and Micron — have pivoted their limited cleanroom space toward higher-margin enterprise products, leaving consumer markets starved.

The consequences are stark. DDR5 RAM prices have surged 171% year-over-year [4]. Contract prices for DRAM are projected to rise 50-60% in Q1 2026 alone [3]. And the GDDR7 memory modules needed for NVIDIA's latest RTX 50-series cards have become so scarce that the company has reportedly paused production of the RTX 5060 and most RTX 50-series GPUs until at least Q3 2026 [5]. The anticipated RTX 50 SUPER refresh has been pushed from Q1 to Q3 2026 [6], while NVIDIA's next-generation RTX 60-series mass production may not begin until 2028 [7].

Source: NVIDIA Quarterly Earnings Reports / Tom's Hardware

Data as of Feb 25, 2026CSV

For gamers, this translates directly into pain at the checkout counter. The RTX 5080, with an MSRP of $1,000, now routinely sells above $1,500. The flagship RTX 5090, listed at $2,000, commands over $3,000 on secondary markets [4]. AMD has notified supply chain partners of a 10% price increase across its entire Radeon product line, with the RX 9070 XT already up 17% over three months [8]. One industry analyst at Acer summed up the outlook bluntly: GPU prices may never return to 2023-2024 levels [9].

Samsung's 8nm Lifeline

Enter the RTX 3060 — or rather, re-enter. The card uses NVIDIA's GA106 chip, manufactured on Samsung's 8nm process node. This is the key detail that makes the revival feasible and, in a perverse way, logical.

NVIDIA's current-generation Ada Lovelace and Blackwell GPUs are fabricated at TSMC on its advanced 4N and 5nm process nodes — the same cutting-edge facilities that produce NVIDIA's massively profitable AI accelerators like the H100 and B200. TSMC capacity is spoken for. Every wafer allocated to a gaming GPU is a wafer that cannot produce an AI chip generating far higher margins. With datacenter revenue reaching $193.74 billion in fiscal 2026 — representing 89.72% of NVIDIA's total revenue — the economic calculus is unambiguous [10].

Samsung's 8nm node, by contrast, has available capacity. The foundry currently runs approximately 30,000 to 40,000 wafer starts per month on the process, roughly 10% of its total foundry output [2]. With Samsung already manufacturing the Nintendo Switch 2's custom SoC on similar process technology, the production lines are warm and the supply chain is intact [1].

Source: BattleforgePC / BestValueGPU / Tom's Hardware GPU Price Tracker

Data as of Mar 9, 2026CSV

Critically, the RTX 3060 uses GDDR6 memory — older, cheaper, and abundantly available precisely because AI companies are not fighting over it. While GDDR7 modules are allocated months in advance to hyperscalers, GDDR6 sits in warehouses, a legacy technology suddenly relevant again.

What $329 Buys You in 2026

The original RTX 3060 launched at an MSRP of $329. It offered 12GB of GDDR6 VRAM, 3,584 CUDA cores, ray tracing support, and DLSS 2.0 — a solid 1080p gaming card that could stretch to 1440p in less demanding titles. Its pricing was quickly obliterated by the crypto boom, with cards reaching nearly $1,300 on secondary markets in June 2021, before settling back to around $290 at its lowest point in September 2024 [11].

In 2026, the RTX 3060's specifications look modest but not irrelevant. According to the February 2026 Steam Hardware Survey, 45.04% of all Steam users still game at 1080p resolution [12]. The card's 12GB of VRAM — ironically more than what NVIDIA initially offered on the RTX 5060's 8GB variant — remains adequate for most modern titles at medium-to-high settings. And NVIDIA has hinted that DLSS 4.5 technology could be extended to older Ampere cards, potentially extending their competitive lifespan [1].

The question of pricing for revived units remains unresolved. Current used RTX 3060 cards sell for approximately $260 on eBay, while new-old-stock units fetch around $339 on Amazon [11]. If NVIDIA prices the revived cards near the original MSRP, they could represent the only sub-$350 discrete GPU option from the company — a remarkable indictment of a market where NVIDIA's cheapest current-generation card, the RTX 5060 8GB, is functionally unavailable.

The AI Tax on Gaming

The RTX 3060 revival is a symptom of a deeper structural shift within NVIDIA. The company's transformation from a gaming-first chipmaker to an AI infrastructure giant is now effectively complete.

In fiscal year 2026, NVIDIA posted total revenue of $215 billion. Of that, gaming contributed $16.04 billion — just 11.45% [10]. While gaming revenue grew 41.34% year-over-year, it is dwarfed by the datacenter business, which generated $193.74 billion, up from $115.19 billion in fiscal 2025 [10]. In Q4 FY2026, gaming revenue actually fell 13% quarter-over-quarter even as overall company revenue grew [10].

The financial incentive structure is clear. Every TSMC wafer that produces an H100 or B200 AI accelerator generates orders of magnitude more revenue than a wafer of RTX 5070 GPUs. NVIDIA has reportedly cut gaming GPU supply by 30-40% for the first half of 2026 compared to the same period in 2025 [6]. No new GeForce gaming GPU is expected until 2027 [6].

Gigabyte CEO Eddie Lin offered a candid explanation of NVIDIA's likely strategy: the company will prioritize products that deliver the highest "gross revenue per gigabyte of GDDR7 memory," effectively rationing scarce memory to whichever product — consumer or enterprise — generates the most profit per chip [13]. For gaming, that calculus is increasingly unfavorable.

The Export Control Dimension

There is another, less discussed factor behind the RTX 3060 revival: U.S. semiconductor export controls. In January 2026, the Bureau of Industry and Security revised its licensing policy, creating a framework for case-by-case review of advanced chip exports to China, including products like the H200 and MI325X [14]. However, shipments to China are capped at 50% of total volume shipped to U.S. customers, and qualifying requires third-party security verification [14].

The RTX 3060, built on Samsung's older 8nm node, falls well below the performance thresholds that trigger export restrictions. NVIDIA can sell these cards freely in China and other markets where advanced chips face regulatory barriers. As one supply chain source told Korean media, NVIDIA is "expanding sales of legacy GPUs" in part as a response to export controls that constrain its ability to ship cutting-edge products to certain markets [2]. The RTX 3060 becomes a workaround: a card that can be sold anywhere, to anyone, without regulatory friction.

Gamers React: Relief or Resignation?

The community response has been mixed. Budget-conscious gamers have welcomed the prospect of any affordable GPU reaching shelves. For the significant number of PC gamers still running older hardware — the RTX 3060 was among the most popular cards on Steam throughout 2024 and 2025 — a fresh production run could extend the useful life of mid-range 1080p gaming systems.

But critics see the revival as a damning signal. "If NVIDIA is truly considering reviving a five-year-old GPU, that suggests the gaming GPU market is under serious strain," one analysis noted, "and raises a bigger question: is NVIDIA willing to sacrifice progress in gaming GPUs to prioritize AI chips where the profits are much higher?" [15]

There are also practical concerns. If production volumes are limited and demand is high, the RTX 3060 could fall victim to the same scalping dynamics that plagued its original launch. The 2021 crypto boom drove RTX 3060 prices to nearly four times MSRP. With today's GPU market already distorted by scarcity, the risk of history repeating is non-trivial [15].

Samsung's Foundry Ambitions

For Samsung, the arrangement carries strategic significance beyond revenue. The company's foundry business has struggled to compete with TSMC at advanced nodes, losing ground on yield rates and attracting fewer marquee customers. But Samsung's mature process nodes — including the 8nm line — offer reliable manufacturing with proven yields and available capacity.

The NVIDIA relationship extends well beyond legacy GPUs. Samsung has joined NVIDIA's NVLink Fusion ecosystem for custom silicon manufacturing [16], is building an AI megafactory powered by 50,000 NVIDIA GPUs [17], and continues to supply HBM memory to NVIDIA alongside its foundry services. The RTX 3060 production restart, while modest in scale, reinforces a deepening partnership that spans memory, manufacturing, and AI infrastructure.

Samsung's 8nm line is estimated to handle 30,000-40,000 wafer starts per month for this production [2]. At typical die yields, this could translate into hundreds of thousands of RTX 3060 GPUs per month — a meaningful, if not transformative, addition to a market desperately short of affordable options.

The Bigger Picture

The RTX 3060 revival is ultimately a story about priorities. NVIDIA — and by extension, the entire semiconductor industry — has made a calculated bet that AI infrastructure is the future, and consumer gaming is an afterthought. The numbers support that bet: datacenter revenue grew 68% year-over-year to nearly $194 billion, while gaming limped along at $16 billion [10].

But games are how most people first encounter a GPU. The gaming community built NVIDIA into a company worth over $3 trillion. And now that community is being asked to make do with five-year-old hardware because the memory it needs is being consumed by machines that generate art, code, and conversation.

Whether the RTX 3060's second life represents a clever stopgap or a quiet abandonment of the market that made NVIDIA famous may depend on what comes next. If new-generation supply recovers by late 2026 and prices normalize, the RTX 3060 revival will be remembered as a footnote — a pragmatic move during an unusual shortage. But if this becomes the new normal — if budget gamers are permanently relegated to recycled silicon while cutting-edge manufacturing serves AI — then March 2026 may be remembered as the moment NVIDIA made its choice explicit.

For now, a five-year-old card is being pulled from retirement. And in a market where the RTX 5090 costs more than many people's monthly rent, even that might qualify as good news.