Not Just GTA 6, Your Next Favourite Game Faces Delays as AI Eats Up GPUs

Looks like we’ll unlock all the secrets of the universe before GTA VI.  

The much-anticipated game may be nearing release, but the global RAM shortage, driven by chip production shifting toward high-bandwidth memory (HBM) for AI data centres, is making game developers nervous. While Sony CFO Lin Tao has reassured the market that the company has locked in the minimum memory required for its end-of-year sales, PlayStation has rolled out a PS5 Pro lottery system in Japan to address console shortages caused by the ongoing RAM crisis.

JP Morgan Global Research estimates that DRAM prices have risen more than 400% from the start of 2024 to the end of 2026.

The gaming industry is now entering an unusual hardware cycle. For some studios in India, GPU availability is crucial for game development. For others, particularly mobile developers, the bigger constraint remains the hardware available to players.

The Crunch

At Ahmedabad-based Funcell Games, which develops multiple 3D as well as interactive mobile games, GPU requirements are growing as its use of 3D content and AI expands.

“GPU availability is increasingly important for us,” CEO Abhishek Malpani tells AIM. The company's games involve UI, video ads, 3D characters, and animation—all of which require GPU computing. However, Malpani says the company's requirements remain below those of AAA console (for high-budget video games) or PC studios.

The collision between AI and non-AI demand is clearer in gaming infrastructure.

“AI workloads are making GPU access harder and more expensive for everyone and not just game developers,” observes Sridhar Muppidi, Co-founder of digital tech company PurpleTalk. 

According to him, gamers may feel the impact through the availability of high-end GeForce RTX chips. The high-density memory modules used in GeForce cards share production lines with NVIDIA’s Blackwell and Hopper architectures.

Muppidi points to strong demand for NVIDIA's B200 and B300 chips, particularly from hyperscale data centres.

“Pure play gaming chips like the GeForce series will continue to be constrained in availability as magnets, rare earths, memory, and all scarce commodities are diverted towards more profitable, high-demand processors for data centres,” he explains. 

A Softer Blow

Many developers say the impact on game development is uneven and cloud infrastructure is helping them avoid a hardware bottleneck.

The impact is clear on infrastructure costs and the pressure to diversify reliance on computing models. At Funcell, rising GPU costs have increased infrastructure expenses, but have not materially changed development plans. The company uses a combination of local hardware, cloud computing, and specialised AI platforms. 

Yet AI's productivity gains are helping offset the added compute costs.

“AI is increasing our GPU requirements, but it has also been a tremendous boon for our business,” Malpani affirms. Funcell is using AI across scripting, coding, UI, 3D asset creation, animation, and video creatives.

The result is a trade-off that could define the next phase of game development. A game that would’ve taken several months, or even years, to develop could potentially be completed in a couple of months with AI-assisted workflows. That means higher GPU consumption, but it does not necessarily translate into higher overall development costs.

Felicity Games, which works across games and consumer applications, sees a similar shift. Founder and CEO Anurag Choudhary says GPU costs are not yet a material concern for traditional mobile game development. The larger expenses remain people, user acquisition, art, content production, and live operations.

Cost pressure becomes notable when studios move beyond occasional AI tool use and start embedding AI into their production pipelines.

“Generating and iterating on images, video, 3D assets, and voice creates far more inference work than occasional tool use,” Choudhary adds. At the same time, smaller, more efficient models are lowering the cost of individual compute units even as overall usage rises.

Renting Capacity

This is where GPUs offered by neoclouds may make or break the deal.

Rather than buying expensive hardware upfront, studios can rent capacity. Felicity uses cloud infrastructure for AI workloads because it can scale compute up or down depending on which experiments prove worth pursuing.

Hyderabad-based mobile gaming studio Metasports Interactive has taken this a step further. It only rents cloud GPUs.

“GPU availability isn't a bottleneck for us today,” Co-founder and CEO Kashyap Reddy notes. “We use on-demand cloud GPUs for training and experimentation, so we can scale compute when needed without maintaining idle hardware.”

The model helps it avoid the cost and maintenance of owning hardware and lets the company scale as needed. For Metasports, the more significant constraint is the GPU inside the player's device. 

Blurred Lines

The boundaries between gaming and AI hardware are beginning to converge as GPU capabilities overlap.

In March 2026, NVIDIA showcased RTX PRO Server infrastructure designed to centralise game-development workflows, including content creation, AI research, and quality assurance, on shared data-centre GPU infrastructure. That suggests that game development may now increasingly resemble an infrastructure service, where studios tap into shared GPU capacity when workloads peak.

Still, pressure on gaming GPUs is likely to persist. NVIDIA CEO Jensen Huang recently commented that the company’s chip sales are likely to double next year, but also observed that AI demand keeps outpacing supply and will for the foreseeable future. 

For now, the immediate consequence may be a change in how studios calculate the cost of experimenting with graphics or storylines. AI makes it possible to generate more assets, concepts, and variations, but each additional iteration carries a compute cost.

“Compute becomes one more input in deciding how much experimentation makes sense,” Choudhary notes. Studios ultimately weigh the total cost of creating a game against expected engagement and revenue rather than looking at GPU spending in isolation.

That may produce an unexpected outcome. Instead of simply making game development more expensive, AI could make it more computationally intensive while simultaneously making other parts of production cheaper and faster.

“The net effect may be more experimentation and more games being made, not simply bigger budgets,” Choudhary adds.