Zack Kass, ex-OpenAI executive: AI will create many winners

Zack Kass, OpenAI's former head of go-to-market, says the artificial intelligence market is still in its early phase, which means the biggest winners are only beginning to emerge. Speaking on the sidelines of the Citi Macro Conference in Hong Kong, he gave the market a simple signal: the time for watching from the sidelines is running out.

The market is not yet carved up

Kass does not see AI as a story about one undisputed champion that will take the whole market. He sees a long cycle of redistributing opportunities. If the technology really is at the start of its journey, big wins have not yet been locked in, and leadership will take shape not only at the level of the models themselves but also in applications, services, infrastructure and business processes. For the market this is an important signal: the window to enter has not closed, and a decision to "wait until everything settles" may turn out to be riskier than a careful launch now.

"AI will create many winners."

The remark came from someone who spent a long time taking AI technology to market commercially and saw how companies make decisions not in theory but in a real operating environment. So the emphasis is not on a catchy slogan but on applied logic. While some argue over whether the sector is overvalued, others are learning to build models into everyday work, testing scenarios and collecting first results. At this stage, the winners are more often not those who announce their strategy the loudest, but those who turn the technology into a measurable result faster.

There will be more than one winner

What matters in Kass's words is the construction itself: in his view, AI will not necessarily create a single new monopolist, but it can produce a broad group of strong players. This resembles earlier technology shifts, where the winners were not only the developers of the base platform but also those who packaged it into a convenient product better than others, embedded it in an industry context or radically improved the economics of a service. In other words, value will accumulate along the whole chain, not in one narrow layer.

For business, this points to several obvious areas of growth:

• launching new AI features in existing products
• speeding up sales and marketing through automation
• reducing the load on support and the back office
• building industry-specific solutions with a clear ROI
• testing hypotheses quickly without long development cycles

A more pragmatic conclusion follows from this logic: the winner may be not only the company that builds its own model but also the one that knows how to apply someone else's correctly. For many companies this is even the more realistic path. Not everyone needs their own foundational AI stack, but almost every business benefits from understanding where a model improves conversion, shortens an operation, raises customer retention or opens a new service format. This is where most of the commercial value will be decided.

What businesses should do now

The main practical takeaway from Kass's position is simple: the observation period is ending. If the market really is early, companies still have a chance to take a strong position without huge budgets, but they need to move through concrete use cases, not an abstract "AI strategy" on slides. The first step is usually to pick an area with a clear metric where the effect shows up quickly: customer response time, cost per lead, document preparation time, ticket classification accuracy or team productivity.

The next step is execution discipline. Winners of the AI cycle will most likely differ not only in access to models but in the quality of implementation: data, processes, interfaces, integration into the working loop and a culture of experimentation. It is no longer enough for a business to buy a subscription to a trendy tool and declare that transformation has begun. It needs owners for each initiative, short iterations, clear KPIs and a willingness to cut what does not deliver results. An early market is good because it gives a head start to those who learn fast.

What this means

Kass's thesis is useful because it lowers the temperature of a false choice between "everything is already carved up" and "we must urgently build our own model." Most likely, the main fight in the coming years will be over the quality of implementation. For business this is good news: there are many forms of winning in AI, and some of them are available right now.