Library/Sam Altman’s Attack, Amazon vs. Starlink, and What Opus 4.7 Actually Means | #248

Peter H. Diamandis

Sam Altman’s Attack, Amazon vs. Starlink, and What Opus 4.7 Actually Means | #248

Key Themes
Claude Opus 4.7AI benchmarksAI labor impactAI backlashdata center politicssatellite competitionAI efficiencytranshumanism
2h 11mApr 18, 2026
Summary

AI progress, social backlash, and the next wave of human-machine transformation

This episode moves from Claude Opus 4.7 and Stanford’s AI Index to the labor effects of AI, public backlash against AI leaders, data center politics, satellite connectivity competition, and the possibility of AI-mediated organizations and enhanced humans. The hosts repeatedly return to a central question: how quickly AI capability is rising, and what social, economic, and institutional systems will bend first under that pressure.

1
Spectrum rights can be more valuable than legacy satellite hardware in direct-to-phone communications.

The Amazon–Globalstar discussion repeatedly emphasizes that the strategic asset is globally coordinated spectrum, not the older satellite fleet itself.

2
AI infrastructure deployment can face meaningful non-technical risk from local and state opposition.

The data-center segment shows that politics, permitting, and voter backlash can delay or redirect buildout even when demand is strong.

3
AI efficiency gains may expand demand rather than shrink it, so lower compute cost is not always bearish for suppliers.

The TurboQuant discussion argues that efficiency improvements can increase deployment, usage, and total demand for memory and devices.

4
Narrative-driven AI pivots can trigger sharp market reactions even before operating fundamentals change.

The Allbirds/Newbird AI example is presented as a case where an AI rebrand can generate a powerful investor response.

5
Personalized medicine and AI-assisted drug discovery are emerging as high-conviction long-duration themes.

The GitLab founder’s cancer story is used to illustrate AI-enabled individualized treatment pathways and faster therapeutic iteration.

Select any chapter text to Deep Dive with AI
01Claude Opus 4.7 Benchmark and Product Shifts

The hosts frame Claude Opus 4.7 as a solid point release rather than a dramatic leap, then focus on what the benchmark data and product changes imply. The discussion emphasizes agentic workflows, reduced low-level controls, larger image inputs, and stronger computer-use performance.

Opus 4.7 is treated as an incremental release rather than a breakthrough.
Anthropic is leaning more on prompting and fewer exposed tuning knobs.
The model appears stronger in coding, computer use, and context handling.
Alignment research and weaker-model supervision are discussed as promising directions.
Image inputs have improved, but image generation is still absent.
02Stanford AI Annual Report Card: Growth, Trust, and Transparency

The episode uses Stanford’s 2026 AI Index to review the state of AI progress, adoption, transparency, and public sentiment. The hosts highlight rapid benchmark gains, rising incidents, and a widening gap between expert enthusiasm and public skepticism.

AI performance is improving quickly, especially in software engineering.
Consumer adoption has risen fast across global markets.
Model transparency is falling even as capabilities rise.
Public trust trails expert optimism by a wide margin.
Governance is portrayed as harder because the technology is advancing faster than reporting cycles.
03AI Is Hitting Young Workers Hardest

The hosts focus on evidence that AI disruption is landing hardest on younger workers, especially early-career software developers. They argue the shift is showing up more as reduced hiring than mass layoffs and encourage young people to move quickly, build companies, and stay geographically flexible.

Employment for younger software developers has fallen sharply since 2024.
Older developers have not seen the same decline, suggesting a generational split.
Hiring freezes may be a bigger mechanism than layoffs.
Entrepreneurship and startup formation are presented as the best responses.
High-density ecosystems like Silicon Valley are framed as an advantage.
04AI backlash goes physical, while states push back on data centers

The conversation turns to violent backlash against AI leadership and to state and local resistance against data center expansion. The hosts argue that trying to pause AI is futile because competitive pressure will simply push development elsewhere, making social license a key constraint.

The episode links AI fear to real-world attacks on Sam Altman’s property.
Local and state opposition to data centers is presented as a growing political risk.
The hosts believe pausing AI would simply hand the advantage to competitors.
Off-world compute is raised as a possible workaround to terrestrial resistance.
Social license is framed as a more important bottleneck than chips or raw compute.
05Workers Training Their AI Replacements

This chapter examines how workers are increasingly training AI systems that may replace them, from annotation jobs to factory motion capture and AI-run storefronts. The hosts then broaden the discussion into organizational design, arguing that AI will compress management layers and reshape how companies operate.

Workers are being paid to teach systems that could automate their own jobs.
Motion tracking and annotation are key methods for capturing tacit knowledge.
AI-run stores are presented as a preview of future micro-enterprises.
The hosts debate labor exploitation versus capitalism’s natural evolution.
AI-mediated management may flatten company hierarchies.
06Amazon and Apple Challenge Starlink in Satellite-to-Phone Connectivity

The discussion analyzes Amazon’s acquisition of Globalstar and Apple’s role in satellite connectivity as a challenge to Starlink’s dominant position. The key strategic takeaway is that spectrum rights, launch cadence, and scale matter more than legacy satellite fleets.

Amazon’s Globalstar acquisition is framed as a spectrum play.
Apple appears to want multiple satellite vendors for redundancy.
Direct-to-phone satellite connectivity is becoming a real product category.
Starlink still dominates scale but must keep accelerating launches.
SpaceX’s future constellation plans underscore the importance of mass deployment.
07AI Efficiency Breakthroughs, Synthetic Religion, and Personal Cancer Moonshots

The hosts move from Google’s TurboQuant efficiency gains to broader implications for cheaper AI deployment, then pivot into AI-generated religion and the possibility of micro sects or cults. The chapter closes with personalized medicine examples and a cautionary look at AI-driven rebrands and market storytelling.

Efficiency gains can lower costs while increasing total usage.
AI is expected to spread across more devices and deployments.
Synthetic religion is treated as a plausible social outcome of AI ubiquity.
AI-assisted personalized medicine could speed up medical discovery and treatment.
Market reactions can be driven as much by AI narratives as by fundamentals.
08Enhanced Games Goes Public

The Enhanced Games are presented as a public-market-ready sports and enhancement venture that combines spectacle, transhumanism, and controversy. The hosts debate whether the event is a serious category-building milestone or a risky celebration of doping and performance enhancement.

The Enhanced Games allow medical enhancement in competition.
The company is said to be going public via reverse merger.
The event is framed as a major transhumanism milestone.
Safety, role models, and medical oversight remain central concerns.
The segment connects enhanced humans to future robotics competitions.
09Human Speciation Forks: Longevity, BCI, Space, and Digital Uploads

The final chapter lays out a framework for humanity’s potential future forks, including longevity escape velocity, brain-computer interfaces, space settlement, and digital uploads. The discussion centers on identity, optionality, and whether these technologies preserve the self or create something new.

Human evolution is framed as becoming intentionally directed.
Longevity, BCI, space settlement, and uploads are presented as future forks.
The speakers debate whether enhancement preserves identity or replaces it.
Optionality is treated as a strategic principle.
The episode ends with outro music and a subscription sign-off.