Skip to main content
  1. Home
  2. Computing
  3. News

Future graphics cards may be twice as fast, but speed comes at a price

Add as a preferred source on Google

As many people still struggle to get their hands on one of Nvidia’s RTX 30-Series graphics cards, preparations for the next generation of GPUs are well underway. The next line of Nvidia graphics, code-named Ada Lovelace, is said to release in 2022.

According to rumors, the new cards may turn out to be twice as fast as the current generation, placing them firmly at the top of any list of the best graphics cards. The dramatic increase in performance comes at a price, though: Nvidia 40-Series graphics cards may also be twice as power-hungry as the current models.

Nvidia GeForce RTX 30 Series
Nvidia

The rumor comes from a well-known leaker on Twitter, @Greymon55, who has a long track record for revealing information about the upcoming releases of both Nvidia and AMD. Previous rumors about the upcoming Nvidia release also come from Greymon55 and Kopite7kimi. Put together, all the leaks are starting to form a fairly clear picture that shows us what to expect from Ada Lovelace GPUs.

Recommended Videos

According to various rumors, GeForce RTX 40-Series GPUs will run on the new AD102 chip, also referred to as ADA GPU. The new cards may offer a massive step-up in terms of the number of CUDA cores. Rumored to have 18,432 CUDA cores, Ada Lovelace GPUs will have almost twice as many as the current generation of Ampere cards.

According to Kopite7kimi, the new RTX GPUs will run on TSMC’s 5nm process node and might have an astounding clock speed of up to 2.2GHz. There may also be a serious jump in teraflops (TFLOPs). Nvidia’s top card, the RTX 3090, currently has 36 TFLOPs of FP32 compute power. Ada Lovelace cards are rumored to pack up to 81 TFLOPs, which is over twice as much as the RTX 3090.

A black EVGA RTX 3090 graphics card with pastel RGB lighting on top.
EVGA

Some users were wondering whether Nvidia will utilize a multi-chip module (MCM) architecture in Ada Lovelace. Nvidia’s upcoming Hopper GPU, made for data centers and working with AI, is supposed to release fairly soon, and it will feature MCM. However, it’s more likely that Nvidia will not use MCM in Ada Lovelace GPUs and will stick to the traditional design.

Although the cards are almost certain to be massive upgrades to the current generation, the power comes at a certain price. Nvidia’s current RTX 3090 is already a power-hungry card, and the rumored RTX 3090 Ti is said to have a TGP of 450W. This means that Ada Lovelace cards may have a TDP of up to 600W. Such high-power requirements will force the use of the best power supply units available.

No release date for Ada Lovelace GPUs has been confirmed yet, but all signs point to 2022. With the ongoing graphics card shortage, it’s possible that many users won’t get their hands on the new cards right away.

Monica J. White
Monica is a computing writer at Digital Trends, focusing on PC hardware. Since joining the team in 2021, Monica has written…
ChatGPT Images 2.5 lets you sketch what you want instead of trying to describe it
ChatGPT images 2.5

OpenAI is rolling out ChatGPT Images 2.5, its newest image generation model, and it brings a solid mix of speed and quality upgrades that should make your AI art sessions a lot less frustrating. People are already generating over 3 billion images every week across ChatGPT and the API, and this update is built to make that process faster and noticeably more accurate.

What makes Images 2.5 different?

Read more
Intel is about to make PC chips even more expensive, and some could vanish altogether
Intel is done competing on price, and its supply chain is warning that your next CPU purchase will cost noticeably more.
MacBook Pro Intel version

Intel's chips are about to get pricier again, and this might not be the last time.

Supply chain sources suggest the company is planning another price hike of roughly 10% starting October 5, 2026. According to a Digitimes report, the company wants to shift from chasing market share through competitive pricing to protecting profit margins. 

Read more
Students who use AI every day may be falling behind, while moderate users pull ahead
New global testing data suggests more AI in the classroom isn't automatically better.
Logo, Text

I assumed more AI access might help students learn faster. However, a new OECD study dispels that misconception for me and everyone else who gave their children a paid version of AI chatbots, thinking it would help them through their educational journey. 

According to PISA 2025 data from 760,000 students in 91 countries, students who use AI daily score meaningfully worse on standardized tests than their peers. The full picture is more complicated than a simple ban would fix.

Read more