How AI changes what gets seen.
“Software is eating the world.”
Marc Andreessen, “Why Software Is Eating the World,” The Wall Street Journal, August 2011.
In the opening weeks of 2026, five companies; Google, Amazon, Microsoft, Meta and Tesla, committed more than $650 billion to AI infrastructure for the year. Data centres, chips, energy contracts, training capacity at industrial scale. Google alone signalled $175–185 billion. Amazon said they would spend $200 billion.
Put those five firms together and their annual AI build out outranks the national budgets of Canada, Australia or Spain, and sits in shouting distance of the UK and India. They are not a country, but their AI CapEx bill is one. This insane level of investment means something structural is taking place.
The AI Narrative is Split
The current AI narrative splits into two camps. One camp says AI will decimate professional services jobs, with AI native startups running at a fraction of conventional headcount. For example:
- Claude Design turns prompts into prototypes, slides, and marketing assets, threaten the design and advertising industries.
- Medvi, a tele health company, created a sensation in March 2026 when they claimed to use AI to automate growth and operations so aggressively that they hit $401 million in 2025 revenue with one person and a 2026 sales projection near $1.8 billion.
However, anyone who has observed technology cycles over the past two decades is entitled to be cautious. The other camp says AI will be like every previous wave; incremental, absorbed, business as usual.
This camp points to the fact that we have been through repeated waves of technology hype where the ‘peak of inflated expectations’ was followed by the ‘trough of disillusionment’. For example:
- The internet was going to kill the travel agency business (travel agents are still alive and kicking).
- Supersonic has been ‘2–5 years away’ for the last 10+ years.
- VTOL and urban air mobility that were supposed to be flying paying passengers across major cities are currently in certification limbo.
In most cases, the technology did not vanish, it just arrived more slowly and more narrowly than the venture capital leaders promised.
The Wrong Question
Perhaps we have been asking the wrong question about AI.
At present, it’s all about how much faster it makes us, how many hours it saves, how much output it generates. It’s all about speed, cost and automation.
However, the real shift sits one level deeper: not how fast AI lets a person work, but the improvement in quality of insight, decision making and problem solving.
Let’s apply that to business aviation: AI does not remove the decisions or the needs for expertise or judgement.
Trip planning, dispatch, charter pricing, KYC and onboarding, customs filings, predictive maintenance, asset valuation, sanctions screening, ownership arrangements, these are decision heavy workflows currently distributed across people of different seniority.
We suggest that the industry forgets about the AI hype and focuses on getting prepared properly. Most of all, we recommend that you don’t believe in the breathless hype.
Here are three suggestions to help you prepare:
- Treat AI as an upgrade in authority, speed and judgement: with AI, the human input can shift to knowing when to accept the output and when to challenge it. AI can enhance expertise, judgment becomes more important, not less.
- Look at incentives and processes, not the tech: All of the research so far points out that when AI is under delivering, the problem is rarely the specific AI model being used. It is misaligned incentives, unclear ownership and processes that have not been upgraded. Applying an existing process or workflow is just scratching the surface.
- Decide what stays human: Some decisions need a person for legal, fiduciary or reputational reasons. Be clear on this decision and write this list explicitly.
In our next instalment we explore how ownership is changing, and what that means for risk, access and control.
