Musk to scale up graphics resolution for Grok to do better than AAA games

Tesla CEO Elon Musk says his artificial intelligence (AI) company xAI plans to improve the capabilities of its latest model, Grok 3, to scale up graphics resolution for developers to build games better than AAA titles.

On Friday, internet personality Mario Nawfal shared a video on social media platform X in which a developer reportedly generated an asteroid-style shooter game simply by describing it in plain English. 

The video originated from Dan Kettlebell, an X employee and co-founder of Kettlebell Krüe collectibles, who mentioned that Grok handled all aspects of the game, including HTML structure, game logic, and score tracking. 

“AI isn’t just writing code. It’s making coding optional,” boasted Nawfal.

In response to Nawfal’s post, Tesla and xAI CEO  Musk said that Grok 3 could easily create the game with basic instructions because it’s “already capable of vastly better NPC dialog.”

 “Now, it’s just a matter of scaling up the graphics resolution for Grok to do better than AAA games,” Musk remarked.

xAI envisions high-resolution gaming through AI upscaling

xAI is reportedly venturing into AI-driven game development with its latest model, Grok 3. The AI platform can generate fully playable browser games from simple instructions with no coding knowledge.

On Thursday, Nawfal shared several examples of games the AI model has created, including a missile-firing game in which enemies are programmed to start slow but speed up after every three kills. 

🚨GROK 3 JUST BUILT A GAME—IN SECONDS

Grok 3 just dropped, and it’s already flexing.

Someone asked for a missile-firing game where enemies start slow but speed up after three kills.

Grok didn’t just answer—it coded the whole thing.

AI casually building arcade games on… https://t.co/IpOrDGswwY pic.twitter.com/FBKdFBojyr

— Mario Nawfal (@MarioNawfal) February 20, 2025

Musk has also claimed that Grok 3 can improve the graphical resolution of games through simple text-based prompts. 

“We’re starting work on integrating dynamically-created high-resolution graphics so that you can make photo-realistic games that evolve differently for every player,” he posted on X.

Still, the billionaire CEO did not make it clear if this applies only to games generated within Grok itself or if the technology could extend to existing games as a separate tool, similar to Nvidia’s Deep Learning Super Sampling (DLSS) or AMD’s FidelityFX Super Resolution.

Gamers laud Nvidia’s DLSS as one of the best options in AI upscaling technology, which enhances graphics by generating extra frames between existing ones, using AI to improve frame rates and smoothness. 

The technology runs on the specialized tensor cores of Nvidia RTX GPUs and is often paired with Nvidia Reflex, which reduces input latency. 

With AMD’s help, Sony developed an Nvidia DLSS equivalent in the form of the Playstation 5 Pro’s Playstation Spectral Super Resolution (PSSR). PSSR can amplify frame rates for certain games from 60 FPS (frames per second), seen in the PS5 standard and slim consoles, to a native 120 FPS. 

xAI hasn’t explicitly said it will create a technology that competes with Nvidia and Sony’s AI upscaling capabilities.

xAI to launch new AI gaming studio

Last Wednesday, Musk and xAI announced plans to establish an AI-driven gaming studio. In a live broadcast on his social platform X, the CEO invited developers to join the new initiative, stating, “We’re launching an AI gaming studio at xAI. If you’re interested in joining us in building AI games, please join xAI.”

The announcement did not specify its focus, development plans, or the types of games it intends to produce. However, a social media post from an xAI employee indicated that the team currently consists of nine members, including Musk himself.

During the broadcast, developers demonstrated Grok 3’s ability to generate a playable version of Tetris using Python, further showcasing the model’s flexibility in game creation.

They also mentioned that xAI’s latest large language model (LLM) has been trained using 100,000 Nvidia H100 GPUs at the Colossus Supercluster in Memphis. Grok 3 also leverages parallelized training techniques to outperform rival AI models, including OpenAI’s GPT-4o, Google’s Gemini-2 Pro, DeepSeek-V3, and Anthropic’s Claude 3.5 Sonnet, across multiple benchmark categories.

Cryptopolitan Academy: Are You Making These Web3 Resume Mistakes? - Find Out Here