KORDUROYTuesday, August 25, 2026 · Run 1 · 8:13 AM ET
Nvidia and SpaceX Aim for Orbit as Anthropic Cedes Ground to Cheaper Rivals
10 stories from 4 sources
TL;DR
- Nvidia and SpaceX are building orbital AI data centers (Starmind) around Nvidia's Vera Rubin racks, chasing a $200B standalone-chip market.
- Anthropic's Fable 5 captured just 11% of corporate AI spend two months after launch as businesses default to cheaper tools like GPT-5.6.
- The US unveiled sweeping new Iran sanctions ('Operation Economic Outcast') targeting oil, crypto, and shipping trade tied to Tehran.
AI Tips2
Use Claude's /eli5 skill to build understanding before diving into details
- Install once: `claude plugin marketplace add anthropics/claude-plugins-community`, then `claude plugin install eli5@claude-community`.
- Before diving into unfamiliar code, an incident, or a tradeoff, ask Claude to /eli5 it first.
- Claude replies with an HTML artifact using big pictures and few words, built for understanding rather than as a final answer.
/eli5 how does this module work /eli5 why did we make this tradeoff /eli5 what caused this incident
Source: The Neuron
Turn your brand identity into a reusable AI design system
- Download Open Design and connect an AI agent such as Codex CLI.
- Click Create New Design System, paste in your website, then Continue; add a logo or a DESIGN.md if it missed anything.
- Select the design system and give it a concrete brief describing what you want built.
Build a 10-second animation explaining our product
Source: The Rundown AI
AI News3
Nvidia and SpaceX are building orbital AI data centers
SpaceX will build its Starmind satellite data centers around Nvidia's Vera Rubin NVL72 rack, targeting the first racks in space by late 2027; Nvidia says each 72-chip rack delivers up to 25x the compute of its older H100. Nvidia's new Vera CPU (88 cores) is also going into SpaceXAI's Grok agent infrastructure as Nvidia chases a $200 billion standalone-CPU market.
Why it mattersSpace-based compute was dismissed as 'ridiculous' by Sam Altman earlier this year; now two credible players are putting real hardware and timelines behind it, worth flagging to clients weighing long-term AI infrastructure bets.
Source: The Neuron, The Rundown AI
Anthropic's flagship model is losing the spending war
Fable 5 captured just 11% of corporate AI spend two months after launch, as businesses defaulted to cheaper tools like OpenAI's GPT-5.6, according to data from 70,000 companies.
Why it mattersA useful data point when clients ask which model to standardize on: price and switching cost are currently beating raw capability in procurement decisions.
Source: The Neuron
Thomson Reuters built its own legal AI model instead of renting one
The two-year effort, built by retooling Alibaba's open-source Qwen and trained on decades of Westlaw and Reuters content, cost $40 million total (the latest training run alone cost $450k); Thomson Reuters says it benchmarked ahead of Claude Opus 4.8, Gemini 3.1 Pro, and GPT-5.5 in internal testing, though only tested in-house so far.
Why it mattersA concrete build-vs-buy case study for any client with a deep proprietary content library: $40M is closer to a year of API bills than a moonshot budget as training costs on strong open base models keep falling.
Source: The Rundown AI
Beyond AI5
The US unveiled sweeping new Iran sanctions
Treasury Secretary Scott Bessent announced 'Operation Economic Outcast,' targeting countries that help Iran trade oil, crypto, gold, aviation, technology, or shipping; the US immediately sanctioned 60 entities, individuals, and vessels, and the UAE already cut trade ties with Iran last week. China buys 80-90% of Iran's oil exports, but enforcement against China is uncertain given a trade truce and a planned Xi Jinping visit to the US next month.
The takeawayIf any client has exposure to Gulf shipping, energy pricing, or China trade flows, this is the kind of geopolitical shift that shows up in cost assumptions within a quarter, not a year.
Conversation starterAre we factoring a Strait of Hormuz risk premium into any energy or logistics cost forecasts right now?
Source: Morning Brew
Trump threatened 50% tariffs on Canadian autos, parts, and steel starting January 1
The move follows the US and Canada failing to reach a deal averting new US tariffs on $20 billion of Canadian goods; Canada has vowed dollar-for-dollar retaliation, and Ontario's premier threatened to cut US access to the province's electricity and critical minerals.
The takeawayDirect relevance for any client in automotive, manufacturing, or cross-border supply chains: a January 1 deadline is close enough to plan around now.
Conversation starterDoes anyone have a client with Canadian manufacturing exposure that needs a tariff contingency plan before January?
Source: Morning Brew
Bitcoin rallied over 22% in a week, closing near $80,000
The move followed the Treasury's bond buyback announcement and reflects what strategists are calling the 'debasement trade': investors rotating out of the dollar into bitcoin and gold as a hedge against growing US debt.
The takeawayNo direct relevance to your work, general awareness only.
Conversation starterAnyone else notice bitcoin creeping back toward $80k while nobody's talking about it?
Source: Morning Brew
NASA-funded researchers 3D-printed cookies out of recycled plastic for astronauts
The process breaks down PET plastic into a carbon-rich liquid, feeds it to engineered yeast that produces protein and fat, then 3D-prints the result into cookies; researchers are still awaiting approval for humans to actually try one.
The takeawayNo direct relevance to your work, general awareness only.
Conversation starterIf NASA can turn plastic bottles into cookies, what's the most 'wait, that's real?' innovation story you've heard this month?
Source: 1440 Daily Digest
Gen Z is embracing deliberately slow messaging apps
Carrier Pidge (virtual pigeons, up to 110 mph, a 0.2% chance of getting 'lost') and Roost (1,000 animal couriers, including a 1 mph snail) have each drawn tens of thousands of users since launching in April, part of a broader 'friction-maxxing' trend alongside dumb phones and social-blocking hardware.
The takeawayA live data point for any conversation about digital fatigue or intentional-friction UX, if a client's product roadmap touches attention or engagement design.
Conversation starterAre we seeing any of this 'friction-maxxing' pushback show up in how younger users react to our own product's engagement features?
Source: Morning Brew
Summarized from The Neuron, The Rundown AI, Morning Brew, 1440 Daily Digest (Aug 25)
Enjoyed this brief? Buy Korduroy a treat
