BREAKING: For the moment, GasBuddy data shows the national average price of gasoline has fallen below $4 per gallon as of 3pm ET, for the first time April 20. In addition, 26 states are seeing average gas prices below $4/gal.

Lunch @PfriemBeer in Hood River

Saturday vibes

AI on my $VCX position: Indeed, Scrib. Though in this case the box appears to contain: - one locked pre-IPO position, - a 50% mark-down truffle, - and a faint note of “portfolio concentration risk” on the finish. Still edible. Just not exactly See’s Candies.

As a result of a US government directive, we are suspending access to Claude Fable 5 for all users. You can continue to use all other Claude models. Here’s what this means for you: Across Claude products, new sessions will run on your selected default model or Opus 4.8, and existing Fable 5 sessions will end with an error. On the Claude Platform, requests to Fable 5 will also return an error. Please update your integrations to other Claude models. We know this is a disruption to your workflows; we appreciate your patience and support.

Leavenworth, WA

Mini retirement party at Icicle Ridge Winery in Leavenworth, WA

Run SkillSpector before installing any new Hermes skill SAFE/LOW (score ≤ 20) = acceptable CAUTION/MEDIUM (score 21–50) = manual review DO NOT INSTALL/HIGH (score ≥ 51 ) = blocked Any high/critical finding is treated as a blocker even if the aggregate score looks softer

After much deliberation; Notion out, Obsidian in. Notion was so easy to sync with the VPS, but the UI was just a non-starter. Obsidian Sync Syncthing is running smooth. Hermes was not into iCloud, although we have used it before. I can't argue with the agent

"Electricity demand growth is led by an ​increase in the commercial sector, which is expected to outpace residential demand in 2026 for the first time on record" reuters.com/business/energy/…

$SPX 3months, hourly...chewing through supports. RSI approaching oversold, but that can stay that way until it isn't...

🚨 $NVDA isn't a chip company anymore. It's the bank, the supplier, AND the customer of the entire AI economy. $40B invested in five months. Here's the map. 🧵 🧠 FOUNDATION MODELS $30B → OpenAI $10B → Anthropic xAI (led ~$20B round) ⚡ COMPUTE & NEOCLOUD up to $2.1B → $IREN $2B → $CRWV $2B → $NBIS 🏭 FOUNDRY & CHIP DESIGN $5B → $INTC $2B → $SNPS 💡 OPTICS & LIGHT $2B → $LITE $2B → $COHR up to $3.2B → $GLW 📡 NETWORKS $1B → $NOK 🔧 CUSTOM SILICON $2B → $MRVL 🤖 AGENTIC SOFTWARE (partnership, not a stake) $NOW → ServiceNow: Apriel models, AI Control Tower, Project Arc 🎤 What Jensen is actually saying: 📉 On the selloff (Seoul, this week): the rout is a "buying opportunity," and the AI buildout is "just beginning." 🔦 On optics (Computex): "Copper has its limits... scale up with copper as long as you can. After that, you scale up with optics." 🤖 On software (Q1 FY27): "Agentic AI has arrived, doing productive work, generating real value, scaling rapidly across industries." 📈 On size: $1T in AI compute demand through 2027, up from his $500B call a year ago. 🌍 On strategy: "We don't pick winners. We support everyone." He's funding the models, the compute, and the supply chain that feeds both, and telling you the light layer is the next bottleneck. The story moved past the GPU. It's the whole factory now. 😎 Long several of these layers. 📈

$LMND's AI edge is real. So many companies love to argue that they do AI better than the competition. Most of the time, evidence is scarce or non-existent. In this case, it's clear. Let's review: 1) Loss Adjustment Expense ratio (cost to handle claims) far better than giant incumbents despite their economies of scale advantage. That lead is growing. 2) Hiking pet premiums at half the sector rate & still outperforming the industry on loss ratios. That's despite a higher new business penalty from faster growth. 3) Strong multi-year loss ratio trends overall. Sharp underwriting enabling granular understanding of customers & actually knowing who to profitably offer lower premiums to. One thing to say it. Another to deliver data that shows it. 4) Lack of headcount growth for 3 years while the book kept explosively growing. 5) Far superior customer service scores despite getting dramatically outspent on tech by competitors. Use their product & compare rapid onboarding & claims mgmt processes vs. others. A lot of companies talk a big AI game. Very few use this technology to actually show real signs of these claims being entirely legitimate. Lemonade is one of those companies.

"Since this is a tool we developed on our own, we were not subject to the commercial product bureaucracy that stymied our early work on detecting and treating sepsis patients" healthcareitnews.com/news/ta…

$CASY continues to be a beast. Picked this up in December via @FrankCappelleri 👍

In the span of 6 months we went from: vibecoders to agentic engineers to loop designers At each evolution, consuming more and more tokens Long anything inference related

$AMZN Trainium3 handles compute-intensive prefill and $CBRS CS-3 handles memory-intensive decode under Amazon Bedrock youtube.com/watch?v=_3IYcMd2…

BREAKING: Apple stock, $AAPL, falls -5% from its high of day after unveiling Siri AI, its biggest AI release ever.