Foresight memos, investment theses, and signals from the futures of AI, DWeb, and emerging tech.
How the rise of camera-and-voice AI glasses reshapes platform power, privacy law, work, accessibility, and the phone's place in daily life.
AI glasses sales are climbing, but the evidence splits on demand depth, the camera and app control. Here is what would settle each question.
Camera glasses are selling in the millions, display models are supply-limited, and a privacy fight is shaping where people can wear them.
Our forecast: acceptance tests, not treaties, will set the standard for human control of drone strikes by 2028, and how that call could fail.
AI terminal guidance is spreading through procurement tests ahead of policy. We trace its effects on budgets, escalation, supply chains and arms control.
Three places where drone warfare's trajectory is set by chip decoupling, commercial AI models and factory robotics rather than by military policy.
Three places where the evidence on military drones and AI pulls both ways, and the near-term events that will show which side is winning.
How jamming, cheap mass and procurement are pushing drones toward autonomy while human-control rules weaken in Washington and stall at the UN.
Our forecast: signed, domain-rooted agents become the admission ticket by 2028, while the first large loss passes every check and lands on a human.
What follows when agent identity binds to domains and named humans while statutes and courts pin every agent action on a person or company.
How agent identity is merging with DNS, agent payments and context files, and why borrowed authority, not forged identity, is the larger risk.
Where the evidence on agent identity conflicts: whether verified agents are safe, who should anchor agent names, and who carries the blame.
Standards, registries and courts are giving AI agents checkable identities. Proving who an agent is still says little about whose orders it follows.
Our forecast: China's insured implants outpace US use by 2030 while Medicare lags, plus how the call fails and which signals would change it.
How approvals and insurance codes for brain implants affect payers, consumer earbuds, neural privacy law and the AI labs entering the field.
Three places where brain-computer interfaces cross other fields: who owns a decoded voice, how China builds industries, and AI's next input channel.
Three open questions in brain-computer interfaces: who pays for implants, how much evidence speech decoding needs, and whether neural privacy laws work.
What the first approved brain implant, two years of at-home speech decoding and a thin patchwork of neural-data laws say about where BCIs go next.
How agent-scoped credentials and signed mandates affect fraud, liability, fees and web pricing, and who stands to gain or lose.
More than 100,000 .si names were registered in 12 days, AI.si is listed at $699,995, and a dedicated auction market has opened. Here is what the prices show and what they do not.
How agent payments are converging with prompt-injection research, open agent standards and paid web content, and the risks that come from the overlap.
Where the evidence on agent payments disagrees: stablecoins or bank deposits, who collects the fees, and who pays when an agent is tricked.
Card networks, banks and processors are converging on agent payments while liability, fraud and fee capture stay unresolved.
We forecast how paid humanoid work splits by bloc and by union status by 2028, and which signals would prove us wrong.
Humanoids are being sold as hourly labor pegged to wages. We trace the effects on utilization, pricing blocs, teleoperation, insurance and unions.
How chip controls, open-weight models and EU AI rules are converging with humanoid robots, and what that opens up or puts at risk.
How cheap Chinese hardware, per-hour robot leases, teleoperation data and trade blocs are shaping who puts humanoids to work, and on what terms.
Three contradictions in open AI: a widening capability gap, export controls that backfire, and local models that get better but pricier to run.
Three open questions on US Internet resilience: which chokepoint breaks first, whether AI helps defenders more, and whether government is shield or threat.
How open-weight AI lines up with machine-speed hacking, a strained power grid and open agent protocols, and what that combination makes possible or risky.
How AI agents, data center power demand and warehouse automation line up with US Internet fragility, and why they make outages last longer.
Our forecast: US procurement will bar Chinese-lineage open weights by 2028, yet Chinese models will still carry most open-model use worldwide.
Our forecast for a multi-day US regional Internet outage by 2029, why power will set the recovery time, and the signals that would change our call.
How Chinese open-weight models becoming the world's default open layer spreads into research, security and policy, and who gains or loses.
What follows when grid power, machine-speed exploits and government standing decide how long a US Internet outage lasts.
Four forces reshaping open-weight AI: China's lead in open models, commodity tokens, costlier local hardware, and a superintelligence race.
Four shifts are concentrating US Internet risk: naming chokepoints, machine-speed attacks, a strained grid and political leverage over providers.
Signals, drivers, and divergent scenarios for the future of voice & audio ai over the next 10 years.
Signals, drivers, and divergent scenarios for the future of superintelligence over the next 10 years.
Signals, drivers, and divergent scenarios for the future of context engineering & .md over the next 10 years.
Signals, drivers, and divergent scenarios for the future of ai & agents over the next 10 years.
Signals, drivers, and divergent scenarios for the future of robotics over the next 10 years.
An emerging cluster of 3 signals points to agent identity standards emerge: what we're seeing and what it could mean.
Signals, drivers, and divergent scenarios for the future of superintelligence & ai agents over the next 10 years.
An emerging cluster of 4 signals points to agent identity & delegation: what we're seeing and what it could mean.
OIR.ai is a rare three-character .ai domain where the Spanish verb 'to hear' meets the AI extension, with deep acronym range for voice, audio, research, and venture-backed technology.
Before .ai made Anguilla rich, the signs were there. Now the same pattern is writing itself in plain text with .md, the file extension every AI agent reads for its instructions.
An analysis of the macroeconomic forces accelerating Artificial Superintelligence and the strategic digital real estate required to navigate the new ecosystem.