Kyle Bennett

Founder & CMO @theArchality • HardOCP Founder • Building the future of efficient AI inference with the Archality Inference Array™

[Kyle Bennett @KyleBennett]

@PcPhilanthropy My daily roadrunner visit.

[Kyle Bennett @KyleBennett]

@BrendanBurkeX @SasaMarinkovic @FrameworkPuter @CORSAIR I do not know. I have not used any of it.

[Kyle Bennett @KyleBennett]

A couple of you have asked. I am running both @FrameworkPuter's Max+ 395 128GB, and @CORSAIR's AI Workstation 300 128GB. Both have been great boxes to work with. These are both tiny boxes too, easy to put almost anywhere. Both have excellent cooling as well. These run fine under heavy GPU/CPU loads. Both solid Agent Computers for @NousResearch's Hermes.

[AMD @AMD]

AMD Ryzen AI Max+ is available now.

16 cores. 32 threads. 40 compute units. 50 TOPS NPU. Up to 128GB unified memory.

Built on x86, it brings breakthrough local AI performance to the enterprise and developer workflows teams already use. No translation layer, no performance penalty, no workflow disruption.

[Kyle Bennett @KyleBennett]

Surely the home Agent Computer is the move. I have my AMD Max+ set up as a server (LM Studio - LM Link) now to all my other computers, attached via Telegram to my phone. Not much reason to be packing around a $5000 laptop if you do not have to.

[Ryan Shrout @ryanshrout]

During the @NVIDIA Q&A with Jensen I had the chance to ask the final question: "With the PC market being low margin and cut throat, why enter it now?"

His answer was compelling. He dismisses margin concerns and focuses on the opportunity to add value and reinvent what we love.

[Kyle Bennett @KyleBennett]

@ryanshrout @nvidia Surely the home Agent Computer is the move. I have my AMD Max+ set up as a server (LM Studio - LM Link) now to all my other computers, attached via Telegram to my phone. Not much reason to be packing around a $5000 laptop if you do not have to.

[Kyle Bennett @KyleBennett]

@HotAisle He speaks highly of you as well. I would like to meet you. Need to all sit down and have a beer soon.

[Kyle Bennett @KyleBennett]

Wanted y'all to meet Liam Coffey. I have been working hand-in-hand with him for a few months now, and he is a stand-up guy, no BS, PC gamer, and computer hardware enthusiast at heart. Pretty damn smart too. Well, he talks over my head all the time, so he is smarter than me, you will have to decide that for yourself.

This Canuck has vision of "what can be" like no one I have ever been exposed to. Give him a follow! He will almost assuredly say something at some point that will blow your mind and make you rethink tech. 🤠

Officially he is Archality's CEO & Chief Architect with 30+ years in silicon design, HPC, and advanced security. Principal architect of BMAC™ and the Archality Inference Array™. AMD Alumni.

[Liam Coffey @_Tiraas]

We have been quietly building something that attacks one of AI’s biggest unsolved problems: the memory wall.

Constantly moving massive model weights between memory and compute is the real driver of high inference cost and power consumption. We think there is a better way.

🧵 1/8

[Kyle Bennett @KyleBennett]

@IanCutress We will certainly be doing that soon. 🤠

[Kyle Bennett @KyleBennett]

I am tremendously excited to be a part of Archality. We are bringing disruption to the AI Inference space.

[The Archality Corporation @thearchality]

We have been quietly building something that attacks one of AI’s biggest unsolved problems: the memory wall.

Constantly moving massive model weights between memory and compute is the real driver of high inference cost and power consumption.

We think there is a better way.

🧵 1/8

[Kyle Bennett @KyleBennett]

@JoshDWalrath

[Kyle Bennett @KyleBennett]

@JoshDWalrath

[Kyle Bennett @KyleBennett]

@scottwasson I see what you did there...and it was awesome. 🤠

[Kyle Bennett @KyleBennett]

@haoailab @ltx_model That is incredible efficiency.

[Kyle Bennett @KyleBennett]

@4l3x_Red Data storage will "never" be alleviated considering how much more of it we produce every day, till we have a breakthrough in technology on that front as well.

[Kyle Bennett @KyleBennett]

I have spent most of my career in the PC hardware world, building communities, reviewing gear, and watching the enthusiast space get squeezed.

For years we funded the tech that made modern computing possible. Then AI showed up and took all the GPUs and high-speed memory.

I am not okay with that.

🧵 1/8

[Kyle Bennett @KyleBennett]

@4l3x_Red With the Archality Inference Array, model weights live structurally adjacent to the compute elements that use them, we are calling these WRAP (Weight Resident AI Processor) tiles.

[Kyle Bennett @KyleBennett]

@liquidai @mlech26l @tugot17 What if you did not need to move the weights at all?

[Kyle Bennett @KyleBennett]

@4l3x_Red You still need CPU and RAM, but in a more traditional server sense as we know it.

[Kyle Bennett @KyleBennett]

8/8 If you are a builder, overclocker, or hardware enthusiast who wants to see GPUs and high-speed memory freed up again, I would love to hear from you.

This is personal for me. Let us take the supply chain back, one builder at a time. Reply or DM me.

#ArchalityInferenceArray #BMAC #PCBuilders #Hardware

[Kyle Bennett @KyleBennett]

7/8 We are in conversations with a select group of strategic and financial partners. If you are an accredited investor with deep experience in AI infrastructure or semiconductors, we welcome private conversations.