AMD is shifting the paradigm from computers that just execute commands to Agentic AI that understands your goals. From the jaw-dropping Strix Halo and Gorgon Halo chips to running 125-billion parameter models locally on Windows, local compute is getting a massive upgrade. Plus, a special appearance from Microsoft’s Pavan Davuluri to announce Project Zenith, and the ultimate reveal of the Threadripper Halo Station — a 96-core local supercomputer.
Read more about it on CNET.com
IFA 2026 Is Full of Weird and Wonderful Tech. We’ve Found the Ones You Can Buy Now
0:00 Introduction to IFA 2026 and the Evolution of Computing
0:25 The Rise of Personal AI and Token Processing Growth
1:10 Shifting the Architecture: Local Compute vs. Cloud
1:51 Defining Personal AI: Local, Private, and Contextual
2:46 Strix Halo and Gorgon Halo: Breaking the Memory Bottleneck
4:18 Lenovo ThinkCenter X powered by Gorgon Halo
4:45 New Class of Laptops: Code Name “Sunday” (HP ZBook)
5:12 Real-time 3D World Generation Demo
6:11 Partnership with Microsoft: Making AI More Accessible
6:40 Windows as an Open Platform for Secure Agentic Experiences
7:14 Introducing Project Zenith for Developers
7:53 The Threadripper Halo Station: A Personal Supercomputer
9:03 Closing Remarks: The Future of Personal AI
Video Credits:
Producer / Video Editor: Vatsal Upadhyay
Thumbnail Designer: Zain bin Awais
Add CNET as a trusted news source
Never miss a deal again! See CNET’s browser extension 👉
Check out CNET’s Amazon Storefront:
Subscribe to CNET on YouTube:
Follow us on TikTok:
Follow us on Instagram:
Follow us on Bluesky:
Like us on Facebook:
CNET’s AI Atlas:
Follow us on X:
Visit CNET.com:
#AMD #IFA2026 #RyzenAI #StrixHalo #Threadripper #ArtificialIntelligence #TechNews #Windows11 #ProjectZenith
@CNET
September 4, 2026 at 1:21 pm
Read more about it on CNET.com
IFA 2026 Is Full of Weird and Wonderful Tech. We’ve Found the Ones You Can Buy Now
@AuroraPhillipsps-v5w
September 4, 2026 at 1:22 pm
Really enjoyed it! Looking forward to more videos from you! 💋🔥
@darinherrick9224
September 4, 2026 at 1:38 pm
I’ve yet to see it do anything useful.
@kcidd12
September 4, 2026 at 2:52 pm
I’m still trying to understand what I just watched lol…I mean the idea of A.I machines sounds good on paper but I have yet to see anything “tangible”
@Strange_err
September 4, 2026 at 1:45 pm
This “personal ai” is real bs. Nvidia bug bit AMD. /s
@BrokenFuse
September 4, 2026 at 2:19 pm
They’re direct competitors. AMD jumped in as soon as Nvidia did.
@Strange_err
September 4, 2026 at 2:25 pm
Yep that’s the usual. Follow the crap is the name of the game unavoidably.
@SMASHSESSION
September 4, 2026 at 3:30 pm
@BrokenFuselol they are family they cant be that competitive
@quincyjones725
September 4, 2026 at 2:27 pm
We aren’t paying for any of that! And it’s just the beginning! We’re just getting started!
@JenHungWANG-s1n
September 4, 2026 at 2:52 pm
Appreciate
@glidverse
September 4, 2026 at 3:19 pm
I was thinking when we will see hbm memory in personal computer and here it is we have 576gb hbm3e. We will see more in future for sure arround 50gb to 100gb is sweet spot I think.
@SOFIA-g2d3f
September 4, 2026 at 3:52 pm
I’ve yet to see it do anything useful. ….
@AchwaqKhalid
September 4, 2026 at 4:16 pm
7:55 I want 😍
@SirusStarTV
September 4, 2026 at 5:46 pm
Do they think masses want the majority of their pc performance and ram usage to be dedicated to AI?
@kwabenaoppong1269
September 4, 2026 at 5:50 pm
Apple M5 max… hold my beer 🍺
@JourneyStand
September 4, 2026 at 5:50 pm
How about you give us affordable and competetive GPUs . Without the crazy price hicks
@igormarcos687
September 4, 2026 at 5:59 pm
There is no personal agentic AI if there is no RAM for personal usage. Shame on AMD for following nvidia and their anticonsumer practices.
@christopherlovett3119
September 4, 2026 at 6:20 pm
Hopefully AMD doesn’t start over hyping their products, acting as if they’re almost sentient, like other tech CEO’s and news media. Open source models running on local machines can help professionals be more productive if used correctly.
@SniperMayer
September 4, 2026 at 6:46 pm
Still nothing on CUDA?
@baa_baa_black_sheep
September 4, 2026 at 6:54 pm
Hardware is not the bottleneck. Software is. Fix that.
@Terrylovesjougurt
September 4, 2026 at 7:05 pm
Where’s the promised Redstone suit? AMD like NV really likes to sell something on promises alone and then either underdeliver or not deliver at all.
@homeape.
September 4, 2026 at 7:56 pm
wow, this was incredibly boring
@sean2val
September 4, 2026 at 9:25 pm
more rubbish