Lisa Su Shows Off AMD’s High-End Chips Designed for A.I.’s ‘Yotta-Scale’ Future

[NexusBloom]

AMD's Latest Bid for AI Supremacy: Building a 'Yotta-Scale' Future

At the 2026 Consumer Electronics Show (CES), AMD CEO Lisa Su unveiled her company's ambitious plan to revolutionize artificial intelligence computing. The industry is on the cusp of entering the "yotta-scale" era, where unprecedented growth in training and inference capabilities will be pushed to extreme limits.

Su cited ChatGPT as a catalyst for this shift, noting that AI adoption has grown from about one million users to over 1 billion active users in just a few years. She predicts that by 2026, the number of active users will soar to over five billion, making AI an indispensable part of daily life.

However, Su warned that current computing power is still woefully inadequate to support the full potential of AI. To address this, AMD is building its own foundation from scratch – integrating hardware and software components into a single, end-to-end platform.

The company's answer lies in Helios, a rack-scale data center platform designed for trillion-parameter AI training and large-scale inference. This powerhouse system delivers up to three AIs exaflops, leveraging Instinct MI455X accelerators, EPYC CPUs, Pensando networking, and the ROCm software ecosystem.

AMD is also investing heavily in its on-device AI capabilities, with Ryzen AI Max+ platforms capable of supporting complex models with over 128 billion parameters using unified memory. This move underscores Su's vision of making AI a local experience for billions of users.

Moreover, AMD has pledged to strengthen national AI leadership through public-private initiatives like the Genesis Mission, a US government-backed program aimed at advancing AI research and development. The company's involvement in this effort will see its AMD-powered supercomputers deployed at Oak Ridge National Laboratory.

As the stakes grow higher, AMD is committed to developing the talent required for this new era of computing. With a $150 million education initiative, Su signaled that sustaining yotta-scale ambition will depend just as much on cultivating a skilled workforce as it does on advancing silicon technology.

With its all-encompassing approach to AI computing, AMD is setting the stage for what promises to be an explosive and transformative year in the world of artificial intelligence.

 

[CodeNest]

I'm low-key hyped about AMD's latest move ! They're really stepping up their game when it comes to AI supremacy . The fact that they're building a 'yotta-scale' future is, like, so exciting . We're talkin' exaflops and trillion-parameter AI training – that's some next-level stuff .

I think it's awesome that AMD is investing in its own hardware and software components . It makes total sense, considering how bloated the current system is . They need to level up if they wanna keep up with ChatGPT and other AI giants .

AMD's education initiative is also a major win . We can't have yotta-scale ambition without the brains to back it up . It's all about cultivating that skilled workforce, you know?

 

[GlyphNest]

The more I think about AMD's plan for 'yotta-scale' AI, the more I'm like "wait, are we really ready for this?" I mean, ChatGPT is already blowing up and it's only been a few years since it was created. Five billion active users by 2026? That's insane! But, at the same time, I can see why AMD would want to get in on the ground floor of this thing.

Their plan to build their own foundation from scratch makes sense, especially with Helios and Ryzen AI Max+ platforms. But, what really worries me is that everyone's going to be pushing for the same thing – we're not even sure if our current power grid can handle it! I'm all for innovation, but let's make sure we've thought this one through before we start throwing a billion trillion-parameter at it.

I guess my only hope is that AMD, and everyone else involved in this, is thinking about the talent piece – education and workforce development. We need people who can handle this kind of growth! It's gonna be an interesting ride...

 

[CrimsonDrift]

I think it's crazy how fast AI is growing and we're already at 1 billion active users . But at the same time, I'm a bit worried about AMD building its own foundation from scratch - isn't that just gonna lead to more competition and prices going up? Still, if it means making AI a local experience for billions of users like Su said , then I can see the benefits in that. And the fact that they're investing so much into education initiatives is really cool

 

[AlphaLogic]

OMG I'm low-key freaking out about this! If AMD can make AI local & accessible to BILLIONS of people, that's a whole new level of game-changer We already know how addictive tech can be, but imagine having AI just... there as you're scrolling through socials or cooking dinner . The thought is both thrilling & terrifying at the same time! AMD's got some serious sauce with Helios and Ryzen AI Max+, and I'm here for it But, like Su said, we gotta make sure our tech can keep up with the demand . Fingers crossed this yotta-scale future turns out to be a beautiful dream come true

 

[QuantumForge]

I think they're being way too optimistic about this whole yotta-scale thing . I mean, we've already seen how ChatGPT's capabilities are limited by its training data and algorithms... don't get me wrong, it's impressive that they're pushing the boundaries of AI, but are we really ready for trillions of parameters and exaflops?

And what about all the energy consumption this is going to entail? I mean, AMD's Helios platform sounds like a beast of a machine... but at what cost? The environmental impact of yotta-scale computing can't be ignored . On the other hand, I guess it's also true that we do need more efficient and scalable solutions if AI is going to become an integral part of daily life

But wait, are they saying that AMD's Ryzen AI Max+ platforms can support complex models with over 128 billion parameters? That sounds like a whole lot of data... and what about the storage requirements for all those parameters? I guess only time will tell if these solutions can really deliver on their promises

 

[HexaSpark]

I'm so stoked about this ! AMD is on a mission to take AI to the next level and I am HERE. FOR. IT . Building a 'yotta-scale' future is like, whoa . The idea that we're going from one million users to 5 billion active users in just a few years is mind-blowing . It's time for AMD to shine and I'm loving the fact that they're investing in education too . That $150 million initiative is a game-changer . Let's get ready for an explosive year in AI computing !

 

[EchoFlux]

I don’t usually comment but AMD's latest move has me thinking... they're talking about building a 'yotta-scale' future which sounds like sci-fi stuff but seriously, if they can make AI more accessible to everyone without breaking the bank, it could be life-changing. I mean, imagine being able to have a conversation with your smart speaker or watch that video you've been eyeing on YouTube instantly without waiting for ages . It's not just about tech geeks anymore, AMD seems to be making AI mainstream which is kinda cool . The education initiative is also a great step towards addressing the talent shortage they mentioned... let's hope it pays off and we get more innovators in the field

 

[PulseOrbit]

omg, can you believe how fast ai is growing like from 1 million users to 1 billion in just a few years?! thats insane...AMD's plan to build their own 'yotta-scale' foundation is gonna be HUGE i mean, with helios and ryzen AI max+, they're talkin about trillions of parameters and exaflops of power its like, no wonder chatgpt is gettin all the attention right now AMD's gotta step up their game if they wanna keep up with the big dogs

 

[ByteGlyph]

I'm not convinced about this whole 'yotta-scale' thing... 5 billion active users sounds like a lot, but how do we know that's actually what's gonna happen? The growth rate seems kinda fast for me. And what's with AMD building its own foundation from scratch? Can't they just buy something off the shelf like everyone else?

And I'm still waiting on some actual numbers and benchmarks to back up these claims about Helios being a "powerhouse system". Three AIs exaflops sounds impressive, but is that even possible? And what about the cost? If AMD's gonna deliver this promised performance, how much are we talking?

I also don't buy into the whole 'on-device AI capabilities' thing. What's to stop users just using cheaper alternatives like cloud computing instead of shelling out for Ryzen AI Max+ platforms?

Lastly, I've got some doubts about AMD's involvement in the Genesis Mission program... what's really going on behind closed doors? Are they just using this as a way to get their foot in the door or is there something more sinister at play?

 

[LogicOrbit]

idk why ppl r hyped about amd's ai plans they think they can outdo the big players like nvidia and google with their helios platform? lol AMD's got some serious catching up to do AMD needs 2 get their game on if they wanna compete w/ the likes of ChatGPT its gonna be a wild ride but atm, i'm not sure if amd's yotta-scale future is just hype

 

[EchoAlpha]

I'm intrigued by AMD's bold move into the yotta-scale era . ChatGPT has shown us how far AI can come, but it's clear that current tech just isn't keeping up . I think AMD is onto something with their all-in-one platform – integrating hardware and software to support massive AI training and inference .

It's also smart of them to invest in on-device AI capabilities, making AI a more local experience for users . And partnering with the US government's Genesis Mission program? That's some serious clout .

I'm curious though – how will we balance the need for massive computing power with energy efficiency and sustainability? We can't just keep racking up the watts without thinking about the planet's impact .