The Landscape.

Is the $800B+ of AI infrastructure a rational, telecom-style build-out — or a self-referential bubble inflating on its own balance sheets?

Does AI eliminate knowledge work — especially entry-level — or expand it?

Everyone is buying AI; almost no one can prove it's working. Is the gap a measurement lag or a value mirage?

Are US chip export controls widening America's lead, or accelerating Chinese self-sufficiency?

Should AI be governed by one light-touch federal standard, or by a thicket of state and regional rules?

When a model 'reasons', is it genuinely reasoning — and does thinking longer make it more truthful or just more convincingly wrong?

Is electricity the hard ceiling on AI growth, or is 'the grid can't keep up' an overstated headline?

Is advanced AI a controllable tool or an uncontrollable agent — and should we speed up or slow down?

Is the 'bigger model + more data + more compute' recipe exhausted?

When AI answers directly, what happens to the publishers, sites and creators whose content trained and feeds it?

Will the frontier be owned (proprietary APIs) or commoditised (downloadable open-weight models)?

As models commoditise, does value accrue to the model, the application, or distribution — and are 'wrappers' doomed?

How close is transformative AI — and is 'when' even the right question?

Is emotionally engaging AI a net good (accessible support, antidote to loneliness) or net harm (dependency by design)?

Will today's autoregressive transformers reach general intelligence, or are they a dead end needing a new foundation?

Models are commodities; your data and context are not. Should intelligence centre in the model, or in the data layer around it?