Why Are GPUs So Expensive? Explained

Graphics cards, more commonly known as GPUs (Graphics Processing Units), have become an indispensable part of any gaming PC build or workstation used for graphics-intensive tasks like video editing, 3D modelling, AI computing, etc.

However, over the last few years, GPU prices have skyrocketed to an all-time high which has made building a PC extremely expensive.

Let’s take a deeper look at the major factors behind the rising prices of GPUs.

Why Are Graphics Cards So Expensive?

GPUs essentially serve as the visual compute engine of computers. They are specialized electronic circuits designed to rapidly process and manipulate memory to accelerate the creation of images, videos, and animations for output displays.

From powering cutting-edge video games to enabling complex design simulations and machine learning models, GPUs handle the bulk of graphical and parallel processing workloads today.

However, this crucial PC component comes at a steep price. Flagship GPUs from NVIDIA and AMD now cost well over $1000, almost double or triple the prices from 5 years back. Mid-range GPUs exceed $500, making building a gaming PC an expensive affair.

So, what has led to the extreme price inflation for GPUs in recent times?

There are a multitude of demand and supply-side factors at play here including:

• Surging popularity of gaming and cryptocurrencies
• The global chip shortage
• Scalping by resellers
• Rising manufacturing and R&D costs
• Gains in performance and capabilities
• Lack of competition
• Tariffs on imports

Let’s analyse each of these factors in detail to understand why graphics cards have become so prohibitively expensive.

Demand and Supply Imbalance

One of the fundamental economic principles is that prices are dictated by the laws of supply and demand.

When demand exceeds supply for a product, its price will naturally go up. The reverse is also true – excess supply over demand causes prices to fall.

Unfortunately for GPU buyers, demand has massively outstripped supply in the last couple of years.

Pandemic Fueled Demand

The COVID-19 pandemic was the catalyst that set off the imbalance between GPU demand and supply. 2020 saw lockdowns and movement restrictions imposed across the world.

With people confined indoors, gaming activity saw a huge boom. According to IDC, gaming hardware sales went up by 55% in the first quarter of 2021 compared to Q1 2020.

Gamers stuck at home either built new gaming PCs or upgraded their GPUs. This massive rise in gaming demand was the first trigger that strained GPU supplies. With production impacted due to the pandemic, supplies couldn’t keep up.

Besides gaming, the pandemic also forced many to work and study from home. With computing needs taken out of offices and into homes, sales of laptops and desktops spiked. All these new PCs also needed GPUs.

Datacenters and cloud services providers also had to rapidly scale their infrastructure to meet the work-from-home driven surge in demand. This also called for bulk procurement of the latest GPUs.

Cryptocurrency Frenzy

Simultaneously, interest in cryptocurrencies like Bitcoin was heating up. Cryptocurrency mining is an intensive computing application that relies heavily on GPUs for mining operations. As crypto prices soared to new highs, miners were keen to amp up their mining rigs and rake in profits.

Hordes of miners swooped onto the GPU market, buying them in bulk quantities much to the dismay of gamers who were left empty-handed. In early 2021, crypto miners were estimated to account for 25% of total GPU sales which is a significant chunk.

This massive demand influx from both gaming and mining segments resulted in supply shortfalls and backorders. GPU prices naturally shot up as desperate buyers were willing to pay exorbitant premiums to secure scarce inventory.

The Global Chip Shortage

The extreme demand was just one part of the equation. What made matters worse was a massive supply-side constraint – the global shortage of semiconductor chips.

GPUs rely heavily on certain key components like graphics processors and memory chips. Both these chips along with other electronics have been in severe shortage since 2020.

Impact on Graphics Processors

Popular GPUs use graphics processors designed by either Nvidia or AMD. For instance, the Nvidia GeForce RTX 3080 uses the GA102 GPU chip which is fabricated by Samsung on its 8nm node process. AMD’s Radeon RX 6800 XT runs on the Navi 21 GPU chip produced by TSMC on its 7nm process.

With every GPU manufacturer vying for these advanced processor chips, foundries like TSMC and Samsung are overwhelmed. Their production capacities are fully booked, with lead times extending to many months. This is severely restricting the availability of GPU processor chips, thereby limiting GPU supply and hiking prices.

Memory Chip Shortage

Memory chips are another key constituent of graphics cards that are facing massive shortages. GPUs use specialized high-speed memory like GDDR6 to deliver smooth performance. However, smartphone and console memory demands have also exploded, leading to dire shortages.

Module packaging delays and component shortages have caused GDDR6 supply to dry up. Nvidia even went so far as to redesign RTX 3080 12GB models to use cheaper GDDR6X memory to mitigate costs. But with the memory chip supply crisis showing no signs of abating, GPU prices will continue to stay high.

Scalpers and Cryptominers

Adding fuel to the fire are exploitative market practices like hoarding and scalping which further feed off the demand-supply imbalance.

Cryptominers often hoard GPUs by buying them up in huge numbers since their mining farms require racks of GPUs. By depriving gamers and other consumers, miners exacerbate the shortages.

Scalpers use bots and automated tools to snap up scarce GPU inventory the moment stocks replenish. They then resell the hoarded cards at ridiculously marked-up prices to desperate buyers who have no other recourse. Their greed leads to additional supply constraints.

These anti-consumer practices became rampant during the pandemic shortages. Unless regulations are enacted to curb such activities, these will continue to be a market reality going forward.

Rising R&D and Manufacturing Costs

To deliver better performance and features, GPUs are getting more complex to design and manufacture. GPU vendors are on a constant technology arms race against each other which comes at a real cost.

Making GPUs More Capable

Modern GPUs have evolved to support advanced capabilities like real-time raytracing, deep learning, and high frame rate support. For instance, Nvidia’s RTX cards contain dedicated ray tracing and tensor cores to enable such computationally heavy workloads. AMD’s RDNA 2 architecture brings raytracing support on consoles and PCs.

Adding these special hardware features requires years of Research and Development (R&D) effort along with silicon redesigns. For example, the Turing architecture powering Nvidia’s RTX 20-series took over 3 years and a $300 million investment. All this upfront R&D cost is included in product pricing.

As expectations rise, GPU vendors have to pack in more cutting-edge features and hardware resources into their designs. Supporting new technologies like PCIe Gen 5 and advanced display outputs calls for further R&D spending. These costs ultimately get passed down as inflated GPU prices.

Pushing Manufacturing Limits

Higher performing GPUs also demand bleeding-edge manufacturing techniques that are pricier. Modern GPUs use silicon lithography processes like Samsung’s 8nm and TSMC’s 7nm to pack billions more transistors into the chip.

Shrinking fabrication node sizes require sophisticated tools and complex multi-step semiconductor production processes. Lower process nodes suffer yield issues as defect rates are higher. All these factors contribute to substantially higher wafer and testing costs, which can double with every process shrink.

Add to this other innovations like faster GDDR6X memory, premium materials like the die-cast aluminum frame in RTX 3090 Ti, advanced power delivery systems and thermal solutions. All these add to the BOM (Bill of Materials) costs of manufacturing modern GPUs.

Lack of Competition

One of the main reasons why dominant GPU vendors like Nvidia can dictate prices is the relative lack of competition in this market. The discrete graphics card space has been dominated by just two players – Nvidia and AMD, with Nvidia holding over 80% market share.

With no other vendor offering high-performance alternatives, there is no pricing pressure on the duopoly to lower prices. Lack of competition enables them to release marginally improved GPUs generations after generation while steadily increasing prices.

The situation may change when Intel enters the fray in 2022 with its Arc discrete GPU lineup. But realistically, it may take Intel several generations to become competitive enough to threaten AMD or Nvidia’s market stronghold. Till such time, buyers will be at the mercy of the two incumbents dictating whatever premium prices they choose.

Import Duties and Tariffs

Another factor specific to the United States market is the imposition of import duties on Chinese goods under Section 301 of the Trade Act of 1974. These tariffs affect GPU pricing in the US in two ways:

• Tariff on finished GPUs: The US levied a flat 25% tariff on all graphics cards and motherboards imported from China from mid-2019. This instantly increased prices by 25% for American buyers.
• Tariff on GPU components: In late 2020, tariffs were imposed on key components like memory modules, printed circuit boards, and transistors sourced from China. Since GPU vendors rely heavily on China for sourcing components, this further inflated costs.

With the US government leaving these tariffs in place, it has decisively contributed to the marked up pricing of graphics cards in America compared to other regions. Unless the tariffs are removed, this will remain a constant factor keeping US GPU prices high.

US Import Tariff Timeline on China

Improved Performance and Features

It’s easy to blame external factors like crypto miners and tariffs as the root causes of inflated GPU prices. But on the flip side, the rapid improvements in GPU performance and capabilities also justify the increase in prices to an extent.

Modern GPUs are leaps and bounds ahead of previous generations in performance. The Nvidia RTX 3080 delivers well over 2X the frame rates in gaming compared to the GTX 1080 released just 5 years earlier. AMD’s RX 6950 XT offers over 50% higher 4K gaming performance versus the RX 5700 XT.

Besides gaming, GPUs have picked up capabilities like real-time ray tracing and AI acceleration using tensor cores. High-speed interfaces like PCIe Gen 4 allow next-gen SSDs to minimize bottlenecks.

Support for advanced display standards (HDMI 2.1) and higher resolutions are built-in. All these collectively take GPUs from being just gaming cards to multi-faceted computing platforms.

So while these substantial leaps in GPU capabilities have their costs, customers are getting more value today. Ultimately you pay for the performance you desire. With GPUs edging closer towards photorealism and realism in gaming, performance comes at a premium.

Brand Value

PC gaming has long had an enthusiast community willing to pay a premium for the best-in-class hardware. Brand loyalty also plays a part in the value perception.

Nvidia has built a reputation of being the gold standard in GPUs given their focus on catering primarily to gamers. Their GeForce brand carries a premium brand value which translates into buyers willing to pay more for the same specs.

For instance, the AMD Radeon RX 6900 XT competes directly with Nvidia RTX 3090 in performance, despite being cheaper. However, for many buyers, the GeForce RTX 3090 holds more value being the flagship of Nvidia’s esteemed RTX line. Brand loyalty adds to the price tolerance.

Specialized Components

Modern GPUs comprise hundreds of individual components that make up the complex circuit board. Many of these components are specialized in nature and customized for the extreme needs of graphics cards.

For instance, GPUs use very high speed GDDR6 or GDDR6X memory not found in typical consumer PCs. The voltage regulator modules (VRM) need to be robust and the PCB has to maintain signal integrity for PCIe Gen 4 speeds.

The materials used for cooling fins, thermal pads, mounts and fans/blowers are also specially chosen to withstand high temperatures. Many of these components are manufactured in lower volumes which drives up pricing as economies of scale don’t apply.

This reliance on an ecosystem of high-performance ancillary components and materials contributes to the overall expensive BOM.

Conclusion

In summary, GPUs have become prohibitively expensive due to:

• Unprecedented demand from gaming, crypto, cloud markets
• Global chip shortage crimping supplies
• Scalpers and cryptominers hoarding inventories
• Rising manufacturing and R&D costs
• Lack of competition in the discrete GPU space
• Tariffs imposed on imports from China
• Substantial generational performance gains

Relief is not expected anytime soon either. With Moore’s Law slowing down, developing faster and more capable GPUs will require even higher investments. As long as demand continues to explode from gaming, data centers, AI, enterprise segments, prices will remain inflated.

Unless new competitors can shake up the duopoly in the coming years, Nvidia and AMD will continue to dominate pricing. For gaming enthusiasts, sadly the days of $500 flagship GPUs are well behind us now. Consumers will have to temper expectations and be prepared to shell out a premium for those coveted extra frames per second.

FAQs