AI stocks faltered, exposing a human skills gap

The recent stumble in AI markets is less a collapse than a correction. OpenAI, Palantir and Nvidia’s dip in August (New York Times) showed that investors are re-pricing AI exuberance. 

Some commentators suggested that the companies were overpriced because the impact of AI on the workforce was over-egged. We don't agree. We believe that AI's promise will be gradually unlocked as tomorrow's executives build their expertise. Leaders will gradually understand and act on the technology's potential to transform their companies, from top to bottom.

Looking past the wobble, the long-term trajectory remains relatively clear, at least to us: AI isn’t disappearing, our understanding is evolving. Rather than simply being a product in itself, AI is increasingly recognised as the backbone of the next generation of companies, coursing between all the operational areas of companies of tomorrow.

Smaller teams, bigger leverage

AI is reshaping the labour market from its foundations. Instead of the sprawling workforces that defined the last century, particularly within corporates, AI-first companies are emerging leaner, faster, and more capital-efficient.

Research from McKinsey and others suggests that automation and generative AI could handle 20–30% of the average knowledge worker’s tasks by 2030. Early-stage startups already prove the point: teams of five to ten people can now achieve what once required fifty. Functions like engineering, design, legal review, customer support and marketing are being partly automated, creating a multiplier effect for small groups.

It’s hardly surprising, though, that this first wave of AI has not yet transformed the workforce in a substantial way. The people running companies are still learning these tools themselves, testing, experimenting, and figuring out how to reorganise workflows around them. The impact so far is more a signal of direction than a revolution already realised.

But when this knowledge diffuses and adoption normalises, the effect will be dramatic: smaller teams, with bigger leverage, will be the rule, not the exception.

The skills bottleneck

And here lies the tension. AI may be rewriting the labour market, but the skills pipeline is badly lagging. Today, the number of people globally who can meaningfully develop, audit, and deploy advanced AI systems remains vanishingly small, hence the value of a world-leading AI engineer being the same as an NFL quarterback or a leading goalscorer in the Premier League. 

For individuals, this scarcity translates into extraordinary earnings potential. For businesses, it means hiring can remain somewhat of a bottleneck, even as tools proliferate.

Education’s catch-up problem

If skills are scarce and earnings are outsized, why not teach AI from an early age…? Because education systems don’t adapt overnight. Teachers themselves need years of training, curricula need vetting, and technologies require regulation and stability before they are normalised into classrooms.

That lag explains why many parents turn to private schools or microschools with the resources to move fast. High-fee institutions can market AI-rich curricula as a competitive advantage (Ackman’s Alpha School is a prime example). For public systems, however, widespread adoption requires accreditation, funding, and proof of stability.

Until AI skills become “normalised”, that is, until tools are stable, ethical standards are embedded, and teacher training has caught up, schools will remain consumers of AI, not owners of it. The near-term burden of education will sit with families, entrepreneurs, and private providers. And technologies require regulation and stability before they are normalised into classrooms… The EU’s AI Act can build confidence in solutions and help define the parameters of legal use, but this is not the same as defining a curriculum of what’s needed practically - we spoke with with the lead Architect of the Act and you can read the articles here and here. 

Accreditation as the bridge

This is why accreditation matters. Schools and parents won’t outsource critical learning to unvetted vendors. Any AI academy or curriculum provider that can secure exclusive endorsements, whether from governments, tech giants, or respected education bodies, will hold the keys to scale. Without that trust signal, the market will remain fragmented, and adoption will lag. Youtube and other unregulated learning sources may bridge the gap in the short-term but this is unlikely to be the long-term solution. 

Strategic takeaways

Shift	Implication
AI is enabling smaller, AI-first teams	More companies, more founders, less headcount needed per venture.
Skills remain scarce	Earnings potential for practitioners will remain extraordinary in the near term.
Schools are years behind	Education systems won’t own AI learning until stability and accreditation are in place.
Private institutions and outsourcing will lead	Niche models thrive first; public systems follow only after normalisation.
Accreditation is the bottleneck	Trust and standards unlock systemic adoption.

The future of work is being built by small, AI-first teams, proof that leverage has shifted from manpower to machine power. But education systems, bound by inertia and trust requirements, are years away from making AI literacy a standard part of the curriculum. In the meantime, the opportunity lies in bridging the gap: creating accredited, scalable platforms that can teach the skills today’s labour market demands but tomorrow’s schools aren’t yet ready to deliver.

The paradox is clear: while AI is compressing company size, it is expanding the skills divide. Whoever bridges that gap between the labour market’s needs and education’s inertia will define the next era of growth.