The Optimism Engine
Packy McCormick's 179th Weekly Dose arrives during market turbulence, but the thesis remains unchanged: innovation accelerates regardless of stock prices. This is optimism as a discipline, not a mood. McCormick frames the current moment as a divergence between financial pessimism and technological acceleration — a gap that defines the weekly's entire purpose.
Model Wars Heat Up
The centerpiece is the simultaneous release of Anthropic's Claude Opus 4.6 and OpenAI's GPT-5.3-Codex. McCormick treats this as more than product updates — it's evidence of recursive improvement. Both labs used their own AI agents to build the next generation.
Packy McCormick writes, "Taken together, we found that these new capabilities resulted in powerful acceleration of our research, engineering, and product teams." This is the mechanism fast takeoff believers point to: models so smart they make the next models smarter.
On Opus 4.6, McCormick notes it's "definitely smarter (although thankfully it's still a shitty writer)." The GPT-5.3-Codex iteration shows tangible progress: "I told 5.3 to throw out that trash and make me something that looked better, and it actually did a decent job in one shot."
Anthropic's Super Bowl commercials mocking OpenAI's planned ads add theatrical tension to the technical race. Jordi Hays and others view the ads as deceptive, but McCormick acknowledges they're "super entertaining."
"I don't know what to say other than have fun playing with your new geniuses this weekend."
Critics might note that "shitty writer" models still produce most online content, and design improvements in one-shot prompting remain fragile across different tasks.
Rocks That Think
Eric Jang's essay "As Rocks May Think" provides the intellectual backbone. Jang, VP of AI at 1X Technologies and former Google Brain robotics lead, traces machine reasoning from symbolic logic systems through Bayesian nets to AlphaGo and today's reasoning models.
Packy McCormick writes, "Jang walks through the intellectual lineage of machine reasoning, from symbolic logic systems that collapsed when a single premise was wrong, through Bayesian belief nets that got tripped up in compounding uncertainty, to AlphaGo's breakthrough combination of deductive search and learned intuition, and finally to today's reasoning models, like Opus 4.6 and GPT 5.3."
The practical takeaway: Jang now runs parallel AI research sessions overnight instead of training jobs. He suspects researcher-level compute will soon be widely available, and when it arrives, demand will explode.
The infrastructure numbers are staggering. Google anticipates $85 billion in 2026 CapEx. Amazon's even larger projection sent its stock tumbling. McCormick invokes Lee Kuan Yew's observation that air conditioning changed civilization by making tropics productive — air conditioning consumes 10% of global electricity, data centers less than 1%. If thinking machines deliver similar productivity gains, inference compute demand will be enormous.
Packy McCormick writes, "The sell-off is ugly, but if Jang is right, all of that buildout and much more is going to be put to use." When McCormick asked Claude Opus 4.6 about the selloff, it responded: "if the bottleneck is inference compute, build the data centers. Vertical integration, baby."
Critics might note that CapEx projections assume sustained AI revenue growth that hasn't yet materialized for most enterprises, and 60% of CEOs report AI projects haven't delivered positive ROI.
Biological Computing Enters the Arena
Anduril's AI Grand Prix drone competition received an unexpected entry: a team planning to use cultured mouse brain cells as the AI software. Cortical Labs' CL1 device — $5,000, lab-grown neurons on electrode arrays, kept alive in life-support housing — learned to play Pong in five minutes with 800,000 cells in 2022.
Packy McCormick writes, "At first look, this seems against the spirit of the software-only rules. On second thought, hell yeah." The neurons run on a few watts and learn from far less data than conventional AI.
Critics might note that biological computing remains experimental and scaling beyond Pong or simple drone navigation faces fundamental challenges in stability, reproducibility, and ethical regulation.
Waymo's Gradual, Then Sudden
Waymo's $6 billion raise values the company at $26 billion — the largest private investment ever in autonomous vehicles. The metrics show the "gradually, then suddenly" pattern: 127 million fully autonomous miles, 90% reduction in serious injury crashes versus human drivers, 400,000 rides per week across six US metro areas.
Packy McCormick writes, "Nearly 40,000 Americans died in traffic crashes last year. The leading causes, things like distraction, impairment, fatigue, are all fundamentally human problems. Waymo doesn't have those problems."
The rollout timeline: 20+ additional cities in 2026, including Tokyo and London. McCormick closes with a personal observation: "My kids are never going to get their drivers' licenses, are they?"
Critics might note that Waymo's six-city footprint remains limited compared to the complexity of nationwide deployment, and regulatory hurdles in international markets like Tokyo and London could delay the 2026 timeline.
The Electronaissance
Contrary Capital's Tech Trends Report provides the macro backdrop. AI adoption curves outpace internet growth: OpenEvidence hit 300,000 active prescribers in 11 months (Doximity took 11 years). ChatGPT reaches 800 million weekly active users. Coding AI tools approach $1 billion in ARR.
Packy McCormick writes, "AI companies are reaching revenue milestones 37% faster than traditional SaaS companies did."
Energy projections: 35-50% US electricity growth by 2040, $1.3 trillion AI-related CapEx by 2027, $1-5 trillion global data center spending by 2030. Wind and solar are the fastest-growing energy sources. US fab capacity projected to grow 203% from 2022 to 2032.
Frontier developments: lunar data storage, underwater data centers with 8x fewer hardware failures, Space Force planning a 100kW nuclear reactor on the moon by decade's end.
Packy McCormick writes, "We are living in a sci-fi novel. What a time to be alive."
Critics might note that infrastructure bottlenecks — aging grids, water stress around data centers, supply chain constraints — could compress these timelines significantly.
Bottom Line
McCormick's optimism rests on measurable acceleration: model capabilities, autonomous vehicle safety, energy infrastructure, adoption curves. The counterweight is equally measurable: CapEx without proven ROI, biological computing still experimental, regulatory friction unaddressed. The verdict: optimism is justified when tied to specific metrics, but the gap between infrastructure investment and enterprise returns remains the critical uncertainty. This weekly dose works because it documents the buildout, not because it guarantees outcomes.