NVIDIA Computex 2024 Recap : Rubin GPU Architecture Unveiled Along With Huge RTX AI PC Developments

NVIDIA’s Computex 2024 keynote, delivered by CEO Jensen Huang, has set the high threshold for the event in Taiwan. Here is what the event mainly covered, focusing on the next phase for AI.

NVIDIA’s CEO Jensen Huang Pulls Off An “AI Masterstroke” In This Year’s Computex, Announces Major AI Developments For Consumers & Clients

For starters, NVIDIA’s keynote started a bit slow with a general talk on where the entire industry was headed, with the firm revealing the important announcements in the latter part of Jensen’s speech, but who are we to argue about?

Team Green had put up a great show, with the firm’s CEO being the primary highlight, especially since Jensen, this time, had managed to attract the spotlight from top tech executives from Taiwan. In his presentation at the beginning, Jensen reiterated the country’s importance for the firm’s hold over the AI markets and claimed that the region is shaping up as a vital player in the industry’s future.

NVIDIA & Ambitions Of 100 Trillion Dollar IT Industry Fueled By Massive Compute Performance

First, Jensen Huang showed NVIDIA’s success in providing the computing power necessary for advancements across different technologies. Jensen showcased the phenomenon of “computing inflation,” where he claimed that with increasing data and parameters, the necessity for adequate power has risen in parallel as well, which is why it has become necessary for companies like NVIDIA to come up with solutions to cater to the demand from the markets. “A new computing age is starting” was the primary motive through which Jensen managed to attract attention.

Apart from computing innovations, Jensen believes that with time, this AI revolution can take the industry to a $100 trillion valuation. In the midst of it lies NVIDIA, which has managed to gain the attention of every major firm involved in the AI race, thanks to its robust platform and exceptional hardware capabilities. Not to mention the “CEO math” instance where Jensen Huang claimed that “the more you buy, the more you save” when talking about in the context of AI computing. Now, on to the mainstream developments.

NVIDIA RTX AI PC Platform Sees New Heights, Courtesy of Massive Developments & Improvements

Everyone expected developments relevant to the new “AI PC” narrative that has entered the computing markets. NVIDIA exceeded our expectations, as the firm revealed a lot of content that was targeted towards heavily competing in this particular segment. To start, the firm showcased the progress it has made when it comes to the AI performance its GPUs have managed to output, and now, the RTX 40 series desktop GPUs scale from 242 TOPS at the entry-level and up to 1321 TOPS at the high end, marking a massive improvement. When you compare these numbers to modern-day NPUs, the gap is massive.

Moreover, NVIDIA announced that Microsoft’s Windows Copilot Runtime will get the “RTX treatment” in light of the recent Copilot hype in the AI PC landscape. NVIDIA states that RTX GPUs will accelerate these new AI capabilities, providing fast and responsive AI experiences on Windows-powered devices, which means that modern-day AI PCs are expected to see massive boosts in AI performance simply due to the extensive support Team Green has managed to provide to manufacturers and consumers as well.

Team Green had some plans for developers as well, as the firm announced the NVIDIA RTX AI Toolkit, which aids developers in building AI-specific applications and will include a suite of tools and SDKs for model customization (QLoRa), optimization (TensorRT Model Optimizer), and deployment (TensorRT Cloud) on RTX AI PCs. With the new RTX AI Toolkit, developers can deploy their models 4x quicker and in 3x smaller packages, ultimately ensuring the optimal development experience. Along with that, the new NVIDIA AI Inference Manager (AIM) SDK was revealed as well, yet again focusing on the development side of things.

The firm also provided an update on NVIDIA’s ACE platform and the debut of dedicated NIMs, inference microservices explicitly designed for edge AI for ACE models. RTX AI PC developments showcased by NVIDIA proved that this is the future of computing indeed, and with that, Team Green is yet again ready to capitalize on the hype.

NVIDIA & Gamers: Announcing The Project G-Assist Along With The New “SFF-Ready” GeForce Guidelines

What’s an NVIDIA keynote without having something for gamers? Well, despite the massive AI boom, Team Green didn’t leave the gaming community at all, as the firm unveiled a new personal RTX assistant called the Project G-Assist. What started as a mere “prank” back in 2017, Team Green did surprise us with the announcement, as Project G-Assist is all about your AI gaming companion. The RTX-powered AI Assistant will take input from the user (voice/text), the gameplay screen, and the game app itself, and through a framework, it harnesses the power of AI to provide gamers with the required information.

This includes getting help in completing in-game missions, asking about the in-game latencies, and even getting suggestions on the best settings for your title. Instead of going through tutorials over the Internet, a simple prompt given to the Project G-Assist will get your job done, and it’s indeed a marvel for plenty of newbies out there, looking to step into the PCMR with little to no background knowledge.

Apart from G-Assist, NVIDIA unveiled a dedicated “SFF-Ready Enthusiast GeForce Card” guidelines targeted at the compact computing segment. With the rise of mini-PCs and compact builds, system integrators saw difficulties in utilizing modern-day components simply because they were not ideal in terms of dimensions in many cases.

The firm laid out dimensions of 304x155x50mm for a graphics card to be compliant with the new standard. The cards under the program shouldn’t exceed 2.5 slots which is the sweet spot for smaller PC builds.

NVIDIA’s Next-Gen Rubin Architecture Unveiled, Bumped Up Memory & Specifications

To top everything off in the end, NVIDIA revealed their next GPU architecture codenamed Rubin which is named after American astronomer, Vera Rubin. This “Rubin” nomenclature was reported by us a long time ago, but Jensen decided to end everything with a shocker this time. Before we talk about Rubin, Team Green disclosed plans to release a supercharged version of its Blackwell GPUs, that would feature higher memory stacks, and performance as well.

Now onto the Rubin architecture, the R100 AI GPU is expected to witness mass production by Q4 2025, with a market debut expected somewhere in 2026. There’s also a “Rubin Ultra” in plans as well, which will be released by 2027. In terms of the expected specifications, NVIDIA’s Rubin R100 GPUs will use a 4x reticle design (versus 3.3x of Blackwell) and will be made using the TSMC CoWoS-L packaging technology on the N3 process node and utilizing the next-gen HBM4 memory type. The performance figures are about to be massive with this one.

Well, that’s the end of the recap for NVIDIA’s Computex 2024 keynote, and to sum it all up, Team Green has clearly shown us why and how will they manage to sustain their dominance over the AI markets, and with the plans the firm has laid out, it will be interesting to see how competitors manage to compete with NVIDIA in the upcoming years.