Ben Thompson

speaker

42,103 appearances 120 recordings 2 series first heard Sep 2024 last heard 21 Aug

Ben Thompson’s voice in public audio — every appearance, attributed to the second.

Trend

recordings per month · last 12 months

Recordings per month over the last 12 months — 39 in all, peaking in Mar 2026 with 5.

Appearances

newest first · ▶ plays the moment

Sharp Tech with Ben Thompson · (Preview) Inference in the Agentic Future, xAI Is Two Companies in One, Q&A on Elon’s Lawsuit, Intel, Apple · 15 May 2026 podcast

And particularly when for these agentic workflows, at least right now, and I I think there will be breakthrough. podcast-host

There will be algorithm breakthroughs, there will be architectural breakthroughs. podcast-host

But for now, a lot of these agentic workloads, the real limiter is memory. podcast-host

Is you it's this KV cash issue. podcast-host

It's pulling in all this context, it's remembering state. podcast-host

And if you want these sorts of things that aren't just useful for a task you define right now, but can be spun up like suddenly something comes up seven, you know, in a week. podcast-host

spun back up and it has all the right context and it knows what needs to be done and it executes a job and then it goes back to sleep or whatever it might be. podcast-host

All that stuff needs memory. podcast-host

And and the memory question is also interesting because everything's been about HBM, high bandwidth memory. podcast-host

The reason we want high bandwidth memory is because we want what do you think we want from high bandwidth memory? podcast-host

want high bandwidth, um is the is the answer. podcast-host

So podcast-host

Well podcast-host

No, you're totally right, actually. podcast-host

But because the reason we want we need it for training is we everything on training is like we it's this highly distributed problem where we want the GPUs to do these calculations super fast. podcast-host

We want to keep the GPUs fed. podcast-host

And so it's like this multivariate problem that NVIDIA has solved way better than everyone else. podcast-host

It it's the not just the fast processors, it's also loading them up with podcast-host

Tons and tons of HBM, uh high bandwidth memory, but then also developing all this crazy networking to tie all this stuff together. podcast-host

So it's not just like you have chips acting as one system, or but you have fleets of them, tens of thousands of them acting at it acting as sort of one chip. podcast-host

Showing 3441–3460 of 42,103 · page 173 of 2106