GPUs are aging like fine wine, with rental rates up across the board since January 1st Approximate change YTD: - H200: +160% - B200: +120% - RTX PRO 6000: +65% - H100: +35% Even the 6-year-old A100 rents for more than it did at the start of the year Charts of the Week:

Moonshot - on purpose - reduces cyber capability in open weights Kimi K3.1 and ensured much lower performance on Exploitbench (74.7 compared to 100 of GPT-6 Astra) & ExploitGym (20.7 compared to 42.7 of GPT-6 Astra) than frontier models. They did this by filtering out datasets, as these capabilities were 'emerging' by themselves & not by RLing on those type of tasks. This is a better approach rather than releasing a very cyber capable model with 'safeguards' that can be easily removed. This is the rigth way, with cyber capable models handed to only responsible parties and put in open domain slowly. But, at the same time frontier labs need to expand access for cyber capabilities to the responsible parties (to their credit - they are doing it). Moonshot - to their credit - has done gentlemanly thing. This is why respectful dialogue is important. But this also tells you China has as Cyber capable models as capable as with United State agencies - even if they may not be public. Capability score screenshot on next page.. @teortaxesTex @zephyr_z9 @AGraylin

That is known trick - well documented (so in training data of these models). Cloud providers specifically guard against it. That said, not all customer enviroments may have these guardrails.

On Terminal-Bench 4.0, which evaluates coding agents on realistic, resource-constrained terminal tasks using strict task-specific tests, K3.1 Pro scores 65.7%, ahead of GPT-6 Astra (58.2%) and narrowly behind Claude Opus 5.5 (66.4%).

my model of flapping airplanes, and ~all "sky-high-target" neolabs besides thinky and ssi, is that they are some combination of extremely successful self deception and con artistry. there are many multi-bil valuation companies out there that have like some hope of doing some successful enterprise sales and consulting, but near zero chance of achieving their stated mission or goals. they seem to exist mostly as a wrapper for very ambitious people to feel like they're able to have more autonomy than they would at a frontier lab and for investors to feel like they maybe didn't completely and utterly fuck up by not getting early allocation to anthropic.

Intel CTO explains why Intel is helping Elon Musk build Terafab, and what a new fab needs besides tools DWARKESH: “I think it’ll be interesting to discuss the partnership you guys have with Elon on Terafab. What’s going on there?” PUSHKAR RANADE (CTO): “Well, I think Elon’s publicly talked about his desire to build a fab.” [Musk clip spliced in]: “We either build the Terafab or we don’t have the chips. And we need the chips, so we’re going to build the Terafab.” RANADE: “Sort of the same thinking. I think he’s seeing just an enormous increase in demand for silicon, for compute and for memory. And I think he came to the conclusion that his own businesses are going to require a lot of silicon. And so he thought maybe he should just do it all on his own. Unlike building a car, for example, this is such an intricate and complicated process. We thought we could work together to help accelerate his bringing up his fab.” DWARKESH: “But because of the bottleneck at the tool level, it seems like the net amount of silicon in the world won’t increase. It’s just that maybe if somebody were to start a new fab, they would get access to guaranteed supply in a way that they wouldn’t have if they’re working with another person’s foundry.” RANADE: “It’s actually not just tools. I mean, a fab does require a lot of power, a lot of water, a lot of consumables. There’s many other input variables as well.” DWARKESH: “But those can be scaled in the way that tools cannot.” RANADE: “Yea. Some more than others, but yes. In one sense, yes, we need a lot more equipment. But it is true, we also need a lot more just raw fab space, raw clean room capacity.”