WEBVTT NOTE The Rundown — nextbig.dev daily audio edition, 2026-07-06 1 00:00:04.500 --> 00:00:18.820 It's Monday, July sixth. The wire spent all week telling you capable AI got cheap and portable. Today two of the biggest builders in the business showed you where the money actually went. 2 00:00:18.820 --> 00:00:33.060 Off the model, and off the chip. Here's the rundown: why Meta's most expensive hardware was sitting idle, why OpenAI's new Codex tier is really a swarm of subagents, and what both say about where the margin lives now. 3 00:00:34.120 --> 00:00:46.760 Start with Meta. It published the engineering behind a problem most AI teams would rather not admit they have: some of the most expensive hardware on the planet, idle, waiting for a file. 4 00:00:46.760 --> 00:01:03.320 The GPUs kept stalling between training steps because the storage underneath them was too slow to serve the data. So Meta rebuilt the whole layer. One fast metadata index, bytes streamed straight to the client, and the data moved into the same region as the chips. 5 00:01:03.320 --> 00:01:16.360 And the numbers are the story. Loading a training set went from about a hundred and fifty minutes down to ten. The worst long-tail job went from eighty-nine hours to a hundred and eighty-two minutes. 6 00:01:16.360 --> 00:01:26.760 An idle GPU is the most expensive line in an AI budget. You pay for the silicon and the power whether it's computing or just standing around. 7 00:01:26.760 --> 00:01:39.640 So the scarce thing here was never the accelerator. It was the plumbing that keeps it busy. That's our briefing's read on the week: the chip stopped being the bottleneck a while ago. 8 00:01:40.540 --> 00:01:56.780 Now go up one layer, into software, same week. Thibault Sottiaux, who runs OpenAI's Codex agent, spent the weekend teasing a new tier called GPT-5.6 Sol Ultra. 9 00:01:56.780 --> 00:02:09.740 And the interesting part isn't a bigger model. Ultra is a harness. It runs a swarm of subagents that actually talk to each other while they work, instead of separate agents that only compare notes at the end. 10 00:02:09.740 --> 00:02:20.940 The reported score is ninety-one point nine percent on Terminal-Bench, against eighty-eight point eight for the base model. Those numbers are unofficial, so hold them loosely. 11 00:02:20.940 --> 00:02:37.340 But here's the tell. The base models all cluster around eighty-eight. OpenAI's own, and Anthropic's Claude Mythos five. So the jump OpenAI is selling comes almost entirely from the orchestration. The model underneath barely moved. 12 00:02:37.340 --> 00:02:51.740 Put the two together and the week has a spine. Meta rebuilt the plumbing below the chip. OpenAI is selling the orchestration above it. The model in the middle is the part anyone can rent. 13 00:02:51.740 --> 00:03:14.620 And one floor down, in the physical world, the same lesson. Nvidia and Intel spent the week talking up American-made chips, and the wafers are real, coming off TSMC's Arizona line. But the packaging, the step that fuses the die to its memory, still happens in Taiwan. Independence runs out at the hardest layer. 14 00:03:17.760 --> 00:03:32.160 To the tape. We're holding the Micron long from yesterday. Every infrastructure story this week routes back to memory, and analysts now put it at around thirty percent of hyperscaler AI spend. 15 00:03:32.160 --> 00:03:46.640 New to the book: we're putting Amkor on watch, low conviction. They're building the advanced-packaging plant in Arizona that's supposed to close that Taiwan gap, but it isn't due until twenty twenty-eight. A name to watch, not a position. 16 00:03:46.640 --> 00:04:01.280 And we're holding Nvidia on watch. Meta's storage fix is the tell we keep tracking: the next lever on cost is getting more work out of the GPUs you already own, and only then buying more. 17 00:04:01.280 --> 00:04:07.040 The tape is the desk's scorecard, not advice. 18 00:04:13.720 --> 00:04:15.800 Quick break — two from the desk. 19 00:04:15.800 --> 00:04:30.680 One we know well: vote dot direct. If you're on an H O A or a board, it runs your elections digitally — secure, verifiable, no paper, no clipboard in the lobby. Point your council to vote dot direct. 20 00:04:30.680 --> 00:04:43.320 And if this is your ten minutes of A I for the day, get the written edition too. The full wire, free, every morning — leave your email at nextbig dot dev. 21 00:04:47.590 --> 00:05:06.510 Our call: within nine months, at least one more hyperscaler publicly reports a double-digit gain in GPU utilization, and credits it to rebuilding the storage or data pipeline that feeds the chips, rather than to buying more of them. 22 00:05:06.510 --> 00:05:17.670 What proves us wrong: if by next April the only way anyone talks about efficiency is still buying more accelerators, and nobody but Meta ties a real utilization number to the plumbing. 23 00:05:17.670 --> 00:05:33.190 The model's cheap now. The chip is close to a spot price. The expensive work moved into the systems around them, and today both Meta and OpenAI showed you exactly where. That's the rundown.