The K-Shaped Economy And The Future Of AI

Dr. TJ Jiang, Chief Executive Officer and Cofounder of AvePoint.

getty​As governments and businesses race to capture AI’s upside, the technology is already reshaping how value is created, who has access to opportunity and which skills command a premium.

This shift is transforming economies around the world. As technology leaders, we have a responsibility to make sure that everyone has a fair shot at making it in this new environment.

I spoke about this earlier this summer in a commencement address at the Singapore University of Technology and Design and share the same sentiment here. To ensure everyone has the same opportunity to succeed in a world that’s dominated by AI, we have to be deliberate about how we build our new economy, always prioritizing trust, safety, opportunity and human dignity above all else. ​

The World Economic Forum’s January report, “Four Futures for Jobs in the New Economy: AI and Talent in 2030,” projects that AI and related technologies will create 170 million new roles while displacing 92 million others by 2030.

That is a net gain of 78 million jobs, but it’s worth emphasizing that workers with AI skills will be the beneficiaries of this new economy.

Workers who possess AI skills already earn a 62% wage premium over those who do not, according to PwC’s 2026 Global AI Jobs Barometer. One year earlier, that premium was 57%. These numbers give hard evidence for what many of us already understand intuitively: AI is remaking the economy at scale in real time.

Economist Peter Atwater popularized the term “K-shaped economy” during the pandemic to describe divergent recovery paths between white-collar workers and service workers. Now, Atwater has argued that one arm of the K shoots upward with AI as well: unprecedented productivity, wealth and agency for those who can use AI to build. The other arm bends downward, leaving the displaced, the unautomated and the disconnected behind. ​​

​As AI becomes more effective and accessible, this temporary pattern is transforming into a permanent design risk. Left unchecked, this divergence creates deep, fractured inequalities that tear at the fabric of our society. ​

But it doesn’t have to be this way. Tech leaders can play an important role in ensuring that the systems they build factor in everyone.

For every workflow you automate, ask what new role, skill or economic activity this creates. If the answer is “none,” you are building for only one arm.

My company, AvePoint, recently released The State of AI 2026 report in partnership with Osterman Research, which surveyed 750 business leaders who had direct responsibility for information management, data security or AI at their organization. The report found that nearly half of employees already rely on AI agents daily or weekly, while nearly nine in 10 organizations have experienced at least one agent-related security incident in the past 12 months.

This shows that adoption is moving quickly, but opportunity and control are not yet moving evenly with it, and that is the real leadership test. To build inclusive, secure AI, you need to create strong data protection and governance frameworks, as workers can only capture the upside of new roles if the systems creating them are trustworthy enough to scale.

Access alone will not prevent a K-shaped outcome. Consumer AI may feel increasingly cheap and available, especially as frontier model providers absorb costs and open-source models lower the price of high-quality AI services. But in the enterprise, unmanaged AI use becomes expensive, risky and uneven.

When AI is costly to scale, companies are more likely to concentrate advanced tools, training and technical support in the teams and use cases with the clearest commercial returns. That leaves others with access in theory, but limited ability to use AI meaningfully, reinforcing the K-shaped divide between those able to capture the productivity gains and those left behind.

Ungoverned AI accelerates the K-divide because only organizations with deep resources can manage the risk. My company’s survey mentioned above found that over 35% of enterprise data is already AI-generated and is expected to reach 42.1% within 12 months.

If that data is poorly governed, the benefits of AI will compound for the few while the risks spread to the many. Trust is not a constraint on innovation. It is the condition that makes innovation accessible to everyone. ​

Every enterprise leader I speak with is planning for the upper arm of the K. Far fewer are designing for the bottom one. That is both a strategic mistake and a moral error.

Companies that build for both arms will build deeper trust, broader adoption and more durable market positions. In a world where trust is already the primary bottleneck for AI adoption, the organizations that earn it across the widest possible base will win.

The leaders who define this era will not be the ones who captured the most value from AI. They will be the ones who created the most opportunity through it. ​

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?