Desktop graphics card shipments hit 12.5 million units in the second quarter of 2026, the strongest quarterly total since Q1 2022, according to Jon Peddie Research (JPR) data reported by Tom’s Hardware on September 11, 2026. The number lands at an odd moment: street prices for add-in boards sit near record highs, GDDR memory costs keep climbing, and NVIDIA just posted its strongest quarterly share of the desktop add-in board (AIB) market in years, at roughly 90%. AMD trails at about 8%, and Intel’s Arc lineup holds close to 2%, per the same JPR figures cited across multiple outlets.
The headline tension is simple to state and harder to explain: shipments rose even as prices rose. Tom’s Hardware frames it as gamers “rushing to beat looming price spikes,” and PC Gamer calls the pattern a defiance of common assumptions about price elasticity in PC hardware. Whatever the exact cause, the data point matters for anyone tracking the desktop GPU and AI chip market in 2026: it reshapes how build guides, upgrade timing, and vendor strategy should be read for the rest of the year.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
Desktop GPU shipments: the Q2 2026 numbers in full
Jon Peddie Research tracks the PC graphics add-in board market on a quarterly basis, and its Q2 2026 report is the source underpinning nearly every outlet’s coverage this week. The topline figure: 12.5 million standalone desktop graphics cards shipped in the April-to-June quarter, up from a weaker Q1. Tom’s Hardware and RamTrend, both citing JPR data directly, put the sequential gain at 5.9% and the year-over-year gain at 7.8%. A few outlets, including TweakTown, cited slightly different rounding (closer to 10% sequential and 6.6% annual), which is typical when secondary outlets recompute percentages from the same base numbers using different reference points. The more conservative, directly sourced figures — 5.9% quarter-over-quarter and 7.8% year-over-year — are the ones worth anchoring to.
Context matters here. JPR’s own Q1 2026 report had shown desktop AIB shipments slipping 0.6% quarter-over-quarter to 12 million units, with the firm projecting a compound annual growth rate through 2029 of negative 3.3%. In other words, the market was supposed to be shrinking. Instead, Q2 2026 broke that trend and delivered the best quarter in four years. That reversal is the real story here: not that shipments went up, but that they went up against both a forecasted decline and a backdrop of higher prices.
HotHardware’s read of the same JPR dataset adds a useful nuance: vendor market share barely moved despite the volume surge. Intel picked up roughly 0.3 percentage points, AMD slipped about 0.16 points, and NVIDIA dipped a fraction of a point. PC Gamer’s version of the numbers describes NVIDIA’s share as dropping “0.01%” quarter over quarter — effectively flat. The takeaway: this was a rising-tide quarter, not a share-shifting one. Every vendor sold more, but the competitive pecking order held.
NVIDIA’s 90% share: how dominant is that, really
A 90% share of any consumer hardware category is a striking number, but it needs to be read narrowly. JPR’s figure applies specifically to desktop add-in boards — standalone graphics cards installed in desktop towers — not the broader PC GPU market that includes notebooks and integrated graphics. On that narrower measure, NVIDIA shipped an estimated 11.25 million desktop discrete GPUs in Q2 2026, AMD shipped roughly 1 million, and Intel shipped several hundred thousand, according to Tom’s Hardware’s breakdown of the JPR data.
That distinction matters because a separate, broader measure tells a more competitive story. TechSpot, citing JPR-linked data, reported that total PC GPU shipments — desktop, notebook, and integrated combined — rose 12.2% in Q2 2026 despite weaker overall PC demand and higher memory costs, and that Intel actually led that broader measure with a 56% share, driven by its integrated graphics footprint across laptops. A separate breakdown of the same trend shows Intel’s total PC GPU share sliding from 61% in Q2 2025 to 56% in Q2 2026, with NVIDIA falling from 24% to 23% and AMD climbing from 14% toward 21% over the same stretch. Put simply: NVIDIA dominates the add-in board market that gamers and PC builders care about, while Intel dominates the much larger integrated-graphics market that ships inside nearly every laptop.
The two datasets are not contradictory — they measure different things. But conflating them, as some casual coverage has done, produces misleading headlines. A 90% figure sounds like near-total dominance; a 56% figure in a different, larger category sounds like a competitive market. Both are true simultaneously, for different segments.
Why shipments rose while prices kept climbing
The most consistent explanation across outlets is pull-forward demand: buyers accelerating planned purchases because they expect prices to keep rising, not fall. RamTrend’s summary of the JPR figures, citing TechPowerUp’s reporting, frames it directly: desktop graphics-card shipments reached the four-year high even as GPU prices rose and GDDR memory costs increased. Igor’s Lab, covering the same quarter from a European vantage point, reported that discrete GPU shipments in the broader PC market rose 12.2% despite what it described as a memory crisis affecting GDDR and system RAM pricing at the same time.
That memory crisis is not incidental to the GPU story — it is arguably the root cause of both the price hikes and the purchase timing. Rising DRAM and NAND costs have squeezed margins across the PC hardware industry throughout 2026, pushing GPU vendors to raise prices on cards that use GDDR memory just as system builders were already absorbing higher DDR5 costs. Tech4Gamers’ coverage of the JPR data noted that NVIDIA’s cards saw the biggest hikes of any vendor’s lineup during this stretch, which tracks with NVIDIA’s dominant share of the high-margin, high-end segment where GDDR7 capacity commands the steepest premiums, a dynamic already visible in modified high-VRAM RTX 5090 variants selling through unofficial channels.
Ongoing GPU pricing trackers cited alongside the JPR shipment data corroborate the pattern: MSRP and street-price statistics for 2026 show sustained upward pressure across nearly every RTX 50-series and Radeon RX 9000-series SKU, even as sell-through volumes stayed strong. That combination — rising prices, rising volumes — is unusual enough that PC Gamer explicitly called it a departure from how PC hardware demand curves normally behave.
Historical context: the last four-year high
Tom’s Hardware anchors the “four-year high” framing to Q1 2022, the last time desktop AIB shipments topped the 12.5 million mark in a single quarter. That period sat near the tail end of the pandemic-era GPU shortage and cryptocurrency mining boom, when demand for graphics cards was driven by a mix of remote-work PC building, gaming demand, and mining rigs competing for the same limited supply. Prices then were also elevated, but for a different reason: scarcity driven by mining demand rather than component-cost inflation.
The 2026 version of high prices plus high shipments has a different engine. There is no mining boom pulling cards off shelves this time. It is memory-cost inflation and pull-forward buying ahead of expected further increases. That is a meaningfully different market dynamic even though the surface-level numbers — high prices, high volumes — look similar to the 2022 peak. It also means the usual playbook of waiting for a supply glut to bring prices down may not apply the same way this cycle, since the bottleneck now sits upstream in memory fabrication rather than in GPU die supply.
NVIDIA vs AMD vs Intel: competitive positioning compared
The three vendors are fighting genuinely different battles right now. NVIDIA’s roughly 90% desktop AIB share reflects near-total control of the high-end and enthusiast gaming segment, reinforced by the RTX 50-series lineup and its lead in AI-adjacent features like DLSS. AMD’s 8% share keeps it as a distant but persistent number two, competing primarily on price-to-performance in the mid-range with its Radeon RX 9000-series cards. Intel’s roughly 2% AIB share looks small in isolation, but the company’s Arc discrete lineup is still a relatively new entrant compared with two incumbents that have decades of driver maturity, partner-board relationships, and enthusiast mindshare behind them.
| Vendor | Q2 2026 desktop AIB share | Estimated units shipped | Change vs Q1 2026 |
|---|---|---|---|
| NVIDIA | ~90% | ~11.25 million | Roughly flat (down ~0.1 pts) |
| AMD | ~8% | ~1 million | Down slightly (~0.16 pts) |
| Intel (Arc) | ~2% | Several hundred thousand | Up slightly (~0.3 pts) |
| Total desktop AIB market | 100% | 12.5 million | Up 5.9% QoQ, 7.8% YoY |
Source: Jon Peddie Research Q2 2026 desktop AIB report, as cited by Tom’s Hardware, HotHardware, PC Gamer, and RamTrend. Figures are rounded as reported by these outlets; JPR does not publish exact vendor unit totals publicly.
When the lens widens to the total PC GPU market — including notebooks and integrated graphics — the competitive picture flips. Intel’s integrated graphics ship inside the majority of laptops sold worldwide, giving the company a 56% share of that broader category as of Q2 2026. AMD’s combined desktop-and-mobile share climbed to around 21% in that measure, its highest point in recent quarters, while NVIDIA’s share of the total PC GPU market sits at roughly 23% — dominant in discrete gaming cards, but a minority player once integrated graphics are counted.
| Metric | Q2 2025 | Q1 2026 | Q2 2026 |
|---|---|---|---|
| Total PC GPU shipments (desktop + notebook + integrated) | Baseline | Declining trend per JPR | Up 12.2% YoY per TechSpot |
| Intel total PC GPU share | 61% | 57% | 56% |
| AMD total PC GPU share | 14% | 20% | 21% |
| NVIDIA total PC GPU share | 24% | 23% | 23% |
| Desktop-only AIB shipments | ~11.6 million (implied) | 12 million | 12.5 million |
Source: Total PC GPU shipment trend data citing JPR-linked research, alongside JPR’s own Q1 2026 desktop-only report. Percentages rounded as published by the reporting outlets.
What this means for PC builders and buyers right now
For anyone shopping for a new desktop GPU in September 2026, the practical read is straightforward: the market has confirmed that prices are trending upward and unlikely to correct in the short term, and the people already buying at these prices are effectively signaling that further increases are more likely than a pullback. That dynamic has already reshaped budget-tier CPU strategy this quarter. AMD’s Ryzen 5 5500F and Ryzen 5 7500, launched September 10, 2026 at $99 and roughly $189 respectively, were positioned explicitly around helping builders economize on memory costs rather than GPU costs, but they reflect the same underlying component-cost pressure rippling through the entire build.
It also reframes how upgrade timing advice should work. The traditional logic — wait for a new generation or a price war to bring costs down — assumes normal supply-and-demand dynamics. When the driver of price increases is upstream memory scarcity rather than GPU die supply or demand softness, waiting does not necessarily pay off the way it has in past cycles. That is part of why JPR’s data shows buyers pulling purchases forward rather than delaying them.
Market impact: what this signals for the rest of 2026
The immediate market signal is that demand elasticity for desktop GPUs has weakened relative to price sensitivity in prior cycles. Vendors reading this data will likely conclude that further price increases carry less near-term shipment risk than previously assumed, which could embolden additional hikes through the back half of 2026, particularly if GDDR and system memory costs continue climbing as they have throughout the year. That is a meaningful strategic signal for NVIDIA, AMD, and Intel alike as they plan pricing for holiday-quarter SKUs and any mid-cycle refreshes, building on price hikes NVIDIA has already pushed through the RTX 50-series lineup earlier in 2026.
It also has knock-on effects for board partners and retailers. A four-year shipment high means channel inventory is moving, which reduces the likelihood of the kind of discounting that typically follows a demand slowdown. Retailers holding stock at current wholesale prices have less incentive to cut margins when sell-through is this strong, which suggests street prices are more likely to hold or climb than soften heading into Q4 2026 and the holiday shopping season.
The memory crisis backdrop, explained
None of this GPU shipment data exists in a vacuum. Throughout 2026, DRAM and NAND flash pricing has climbed steadily as fabrication capacity has struggled to keep pace with demand from AI data center buildouts competing for the same memory supply that consumer GPUs and PCs rely on. GDDR memory, used specifically in graphics cards, has not been insulated from that pressure. Igor’s Lab’s framing of Q2 2026 as occurring despite a memory crisis captures the core tension: GPU vendors are absorbing higher input costs for the memory chips that sit on every graphics card, and those costs are being passed to consumers through higher MSRPs and street prices.
This is structurally different from the 2020-2022 GPU shortage, which was driven by semiconductor fabrication capacity constraints for the GPU dies themselves, compounded by cryptocurrency mining demand. The 2026 pressure point sits one layer up the supply chain, in memory manufacturing, and it is affecting far more than just graphics cards. It is the same dynamic pushing up prices for RAM kits, SSDs, and even smartphones this year.
Competitive strategy: how NVIDIA, AMD and Intel are responding
NVIDIA’s strategy in this environment is largely a continuation of its existing playbook: lean into the high-margin enthusiast and AI-adjacent segments where its 90% AIB share gives it substantial pricing power, and let the broader integrated-graphics market, where it has no presence, be someone else’s battle. AMD’s approach centers on defending its mid-range price-to-performance positioning with the Radeon RX 9000-series, a strategy that has kept its desktop AIB share roughly stable even as total PC GPU share, driven by mobile and integrated wins, has grown to around 21%. Intel’s dual-track approach, a still-small discrete Arc AIB presence paired with a dominant integrated graphics footprint across its CPU lineup, makes it the most differently positioned of the three, competing on volume and ubiquity in laptops rather than on enthusiast desktop mindshare.
None of the three vendors appears positioned to meaningfully disrupt NVIDIA’s desktop AIB dominance in the near term. AMD would need either a significant price advantage or a performance leapfrog to move share meaningfully, and Intel’s Arc lineup remains early enough in its maturity curve that single-digit desktop AIB share is a realistic ceiling for the next few quarters rather than a floor to build from.
How this compares to past GPU market cycles
The 2020-2022 GPU shortage era offers the closest historical parallel, but the mechanics differ in an important way. That period saw shipments constrained by supply, with demand from miners and gamers competing for a limited number of GPU dies coming off constrained fabrication capacity. Prices rose because supply could not meet demand. In 2026, supply of GPU dies themselves is not the constraint. TSMC’s own record revenue results for the period, covered separately by outlets tracking the September 2026 earnings cycle, show the foundry continuing to scale output aggressively. The constraint instead sits in the memory components that attach to those dies, a narrower and arguably more persistent bottleneck since it is tied to a global memory market also being pulled on by AI infrastructure buildouts that show no sign of slowing.
That distinction matters for anyone trying to predict when relief might arrive. GPU die supply constraints have historically resolved within a few quarters as fabrication capacity caught up or demand cooled. Memory supply constraints tied to AI infrastructure demand have a much longer runway, since data center memory demand is not tied to the same boom-bust cycle as cryptocurrency mining was in 2021-2022.
What buyers and builders should watch next
The next data point worth tracking is JPR’s Q3 2026 report, expected in the coming months, which will show whether the Q2 surge was a one-off pull-forward event or the start of a sustained recovery in desktop AIB demand. A second data point worth watching is whether AMD’s Radeon RX 9000-series pricing moves in response to NVIDIA’s continued price strength; historically, AMD has used pricing as its primary lever to gain share when it cannot compete on raw performance at the top end. A third is whether Intel’s Arc lineup sees any meaningful uptick in desktop AIB share as newer graphics silicon filters into more system builds. Even a move from 2% to 3-4% would represent a significant relative gain for the newest of the three vendors.
Predictions: where the GPU shipment market heads next
- Prices likely hold or climb through Q4 2026. With sell-through strong enough to produce a four-year shipment high, neither vendors nor retailers have much incentive to discount heading into the holiday quarter, especially while GDDR and DRAM costs remain elevated.
- NVIDIA’s desktop AIB share stays in the high-80s to low-90s range. Vendor share barely moved in Q2 2026 despite the volume surge, and nothing in AMD’s or Intel’s current desktop roadmap suggests a near-term shift large enough to meaningfully erode that dominance.
- Intel’s integrated-graphics lead narrows further as AMD’s mobile share grows. The trend line shows Intel’s total PC GPU share sliding from 61% to 56% over recent quarters while AMD climbed from 14% to 21%; if that trajectory continues, Intel’s total-market lead could shrink further by early 2027.
- Pull-forward buying could produce a softer quarter later. If a meaningful share of Q2 2026’s volume represents demand pulled forward from future quarters, JPR’s Q3 or Q4 2026 report could show a corresponding dip once that pent-up buying is exhausted.
- Memory costs remain the swing factor for the whole category. As long as DRAM and NAND pricing keeps climbing on AI data center demand, GPU vendors have limited room to reverse recent price hikes, regardless of how shipment volumes trend.
Frequently asked questions
How many desktop graphics cards shipped in Q2 2026?
Jon Peddie Research reported 12.5 million desktop add-in board (AIB) graphics cards shipped in Q2 2026, the highest quarterly total since Q1 2022, as reported by Tom’s Hardware and other outlets on September 11, 2026.
What is NVIDIA’s market share of desktop GPUs in 2026?
NVIDIA holds roughly 90% of the desktop add-in board (AIB) market as of Q2 2026, per JPR data. That figure applies specifically to standalone desktop graphics cards, not the broader PC GPU market that includes laptops and integrated graphics.
Why did GPU shipments rise while prices are also rising?
Outlets including Tom’s Hardware and RamTrend attribute the pattern to buyers pulling forward planned purchases ahead of expected further price increases, driven largely by rising GDDR and DRAM memory costs rather than GPU die shortages.
What is AMD’s share of the desktop GPU market?
AMD holds approximately 8% of the desktop AIB market in Q2 2026, according to JPR figures cited by Tom’s Hardware and PC Gamer, roughly stable compared with the prior quarter.
Does Intel compete in the desktop GPU market?
Intel’s Arc discrete GPU lineup holds around 2% of the desktop AIB market as of Q2 2026. However, Intel leads the much larger total PC GPU market, including laptops and integrated graphics, with roughly 56% share, according to shipment data reported by TechSpot.
Is this the same GPU shortage as 2021-2022?
No. The 2020-2022 shortage was driven by constrained GPU die fabrication capacity combined with cryptocurrency mining demand. The 2026 price pressure is driven primarily by rising GDDR and DRAM memory costs tied to AI data center demand, not GPU die supply.
Will desktop GPU prices come down soon?
Nothing in the current data suggests a near-term correction. Strong sell-through in Q2 2026 reduces retailer incentive to discount, and elevated memory costs continue to pressure vendor pricing heading into Q4 2026.
Where does this GPU shipment data come from?
The figures originate from Jon Peddie Research’s quarterly PC graphics add-in board market report and have been reported by Tom’s Hardware, PC Gamer, HotHardware, Tech4Gamers, TechSpot, RamTrend, and Igor’s Lab, among others, in coverage published September 11-12, 2026.
Related Coverage
- Nvidia DLSS 5 in NBA 2K27: 6 Outlets Confirm 50%+ FPS Hit [2026]
- NVIDIA Gives Away 4 RTX 5090s in NBA 2K27 Contest [2026]
- Xeon 6+ Buried as Nvidia PC Chip Sinks INTC 4.2% [2026]
- Apple A20 Pro Leak: 4.90GHz CPU, 7-Core GPU Beats M3 [2026]
- A20 Pro Goes 2nm: GPU Up 40%, Bandwidth Up 50% [2026]
