Artificial intelligence is rapidly changing how companies build products, structure their operations, and compete for value. Yet adopting AI tools is not the same as achieving meaningful digital transformation. As generative AI becomes increasingly accessible, companies must determine whether their investments create lasting competitive advantages or simply deliver improvements that competitors can easily reproduce.
To examine these changes, we spoke with Yuri Romanenkov, a strategy scholar and practitioner specializing in AI value creation, platform economics, and ecosystem-based competitive advantage.
He is the coauthor of the white paper How to Reap Value from Generative AI, which explores how organizations can move beyond superficial AI applications and generate measurable value through proprietary data, knowledge assets, integration into decision-making, and new AI-enabled business models.
His expertise is also informed by his doctoral research on platform economics, network effects, and platform design, as well as his work at London Business School on generative AI, strategy, and business ecosystems. He currently serves as an Academic Director at London Business School and has spent two decades working with technology companies and their investors, from early-stage ventures to established industry players.
In this interview, he discusses how AI is reshaping software economics, why traditional SaaS pricing is under pressure, and what companies must do to build advantages that competitors cannot easily replicate.
Recognizing Real Digital Transformation
1. You have spent 20 years working with technology companies and the investors backing them, from early-stage ventures to established players. Was there a particular moment that changed how you distinguish genuine digital transformation from technology that merely looks impressive?
The turning point did not come from a single eureka moment. It came from repeatedly observing the same pattern over many years.
Companies would invest millions in impressive frontend applications, enormous data lakes, or the latest technological trend—from big data and Web3 to the metaverse—yet their competitive position and profit margins would remain largely unchanged.
That pattern made the distinction between technology adoption and genuine transformation increasingly clear. Technology may appear impressive when it improves an existing process or adds a polished digital layer to the business. It becomes transformative only when it changes the underlying economics of the organization: how value is created, how unit economics work, or how competition within the industry is structured.
When an investment does not produce a defensible advantage or create ecosystem complementarities that competitors would struggle to reproduce, it is usually an expensive operational improvement rather than a true transformation.
What Platform Economics Reveals About Competition
2. What first drew you to platform economics, and what does it reveal about competition that traditional business strategy often overlooks?
My interest in platform theory and business ecosystems came from a fascination with how digital markets challenge traditional economic assumptions.
Conventional strategy often views competition through a relatively rigid framework: one company competing against another within clearly defined industry boundaries, with each business managing its position along a linear value chain.
Platform economics changes that perspective completely. Competition is no longer limited to managing assets and capabilities within the boundaries of one organization. It increasingly involves coordinating external ecosystems, developing multi-sided network effects, and using complementarities as competitive levers.
Within a platform ecosystem, a company’s most important source of value—or its most serious competitor—may come from an entirely different sector. Control over the customer interface, ecosystem relationships, and data feedback loops can become more important than direct ownership of physical assets.
A traditional pipeline company attempting to compete against a platform without changing its strategic model will almost always be at a disadvantage.
The SaaS Valuation Collapse
3. The so-called “SaaSpocalypse” reportedly erased $2 trillion from software company valuations. Do you see this as a fundamental shift in the software business model, or mainly as an investor overreaction?
It is almost certainly a combination of both.
Major technological shifts create radical uncertainty. The range of possible outcomes becomes less predictable, and it becomes extremely difficult to assign reliable probabilities to those outcomes. That uncertainty naturally produces exaggerated reactions from investors, both positive and negative.
Even after accounting for market overreaction, however, a fundamental change in software business models is clearly taking place. The narratives surrounding that change are often too simplistic.
I do not have a crystal ball, but I find it unlikely that companies will suddenly bring most of their software development in-house. No organization is going to “vibe-code” its entire enterprise resource planning system.
What is changing is that software development has become accessible to a much broader group of people, while developer productivity has already increased substantially. Traditional software advantages, including high switching costs, are now under genuine pressure.
The greatest threat to established software companies may not necessarily come from their customers or even from new AI-native challengers. It may come from existing competitors that use AI more effectively to improve their products, lower their costs, and move more quickly.
Why Per-Seat SaaS Pricing Is Under Pressure
4. SaaS pricing is traditionally based on the number of employees using a product. If AI agents begin performing more of the work, will that pricing model stop working? What will separate companies that adapt from those that fail?
The traditional per-seat pricing model is fundamentally incompatible with an economy increasingly shaped by AI agents.
If one AI agent can perform work previously handled by 10 employees, charging per user effectively penalizes the software provider for delivering greater efficiency. The provider creates more value for the customer but receives less revenue because fewer human users need access to the system.
Companies will probably move toward value-based or consumption-based pricing, although neither approach is simple.
Value-based pricing is attractive in theory but difficult to implement because it can be extremely hard to attribute a specific business outcome to one piece of software. Consumption-based pricing is easier to measure, but it can expose customers to unpredictable and potentially substantial costs.
The deeper strategic question is where a company can build advantage in an agent-driven environment. Which traditional sources of advantage have weakened or disappeared? Where will the new bottlenecks emerge? Can the company realistically control those bottlenecks?
Answering those questions requires honest reflection and considerable courage. Businesses that remain attached to the strategies that made them successful in the past are unlikely to thrive when the competitive environment changes dramatically.
Why Software Is More Exposed to AI Disruption
5. Software companies appear more exposed to AI disruption than industries such as retail, energy, or manufacturing. Why are some sectors easier for AI to disrupt than others?
The difference comes down to a basic economic distinction: the physics of bits versus the physics of atoms.
Software depends primarily on digital information, structured data, and cognitive labor. Generative AI can directly substitute for parts of both the product and the work required to produce it. Code can now assist in writing, testing, and improving other code.
Retail, energy, and manufacturing remain deeply connected to the physical world. These industries face capital requirements, complex supply chains, physical infrastructure, regulatory constraints, and legacy hardware. AI cannot independently manufacture a battery, move a shipping container, or replace an industrial production line.
In these sectors, AI is more likely to function as an operational tailwind. It can help optimize production yields, anticipate equipment failures, improve inventory management, and personalize customer experiences.
Paradoxically, the physical friction within these industries can protect incumbent companies from immediate disruption. AI becomes an important optimization tool rather than an overnight substitute for the entire business.
Why Broad AI Strategies Fail
6. Many companies announce broad AI strategies but struggle to produce meaningful results. What usually goes wrong, and what does a genuine AI advantage look like when competitors cannot easily copy it?
Most broad AI strategies fail because they focus almost entirely on generic productivity improvements and cost reduction.
Companies distribute ChatGPT licenses across the workforce or automate basic customer service interactions, but these initiatives rarely create a sustainable advantage. Our research on generative AI value creation shows that when the same technology is available to everyone, much of its benefit is eventually competed away.
If every company in an industry reduces costs by 15% using the same off-the-shelf AI tools, market prices may decline, margins may narrow, and no individual company necessarily gains a lasting advantage.
A defensible AI advantage does not resemble a conventional technology installation. It requires deeper structural and organizational redesign. At present, only around 22% of firms believe they are on track to achieve truly transformative impact from generative AI.
A sustainable advantage rests on two primary foundations.
The first is the creation of proprietary data loops that competitors and publicly available large language models cannot access.
The second is organizational complementarity: redesigning workflows, decision-making processes, and ecosystem partnerships in a way that creates significant friction for competitors attempting to reproduce the same model.
What Leaders Misunderstand About AI
7. What do founders, executives, and investors misunderstand most about AI today? Has anything about AI adoption changed your own perspective over the past year?
The most significant misunderstanding is the tendency to confuse technological capability with structural business defensibility.
Founders and investors often see an extraordinary AI feature and assume that it automatically represents a strong business model. It does not. When a competitor can reproduce the product over a weekend using the same foundation models or open-source alternatives, the company does not possess a sustainable business advantage. It has only discovered a temporary arbitrage opportunity.
I have also changed my view regarding the speed at which incumbent companies can adapt.
Initially, I expected agile, AI-native startups to dismantle slow-moving corporate incumbents relatively quickly. However, established enterprises possess substantial advantages in scale, distribution, customer relationships, and accumulated data.
The rapid development of open-source AI has also made advanced capabilities available much faster than many anticipated. As a result, access to the technology itself is no longer the primary constraint.
The central bottleneck is organizational inertia. Incumbents that recognize this and adjust their strategies decisively are adapting much faster than many observers expected.
What the Software Industry May Look Like Next
8. Looking three to five years ahead, which of today’s SaaS giants will remain important, and what might the software industry look like after the current AI transition?
The SaaS companies most likely to remain influential are those with exceptionally strong data gravity and deep architectural lock-in.
Microsoft and Oracle will continue to matter because they control foundational enterprise systems of record. Replacing a company’s core data architecture is extraordinarily difficult, risky, and expensive, particularly for a Fortune 500 organization. That remains true even when a new generation of standalone AI tools offers more advanced capabilities.
However, continued relevance does not guarantee that these companies will preserve their historical profit margins.
This transition differs substantially from the shift to cloud computing. During the cloud era, Microsoft, Amazon, and Google operated within a relatively disciplined oligopoly and often focused on different customer groups or market segments.
Today, hyperscalers and major AI laboratories are entering many of the same markets while committing hundreds of billions of dollars in capital expenditure. That intensity of competition is likely to generate substantial pricing pressure and margin compression.
Vertical software providers may respond by adding more service components, industry-specific capabilities, and hardware integrations to establish new sources of defensibility. The market is also likely to consolidate.
The companies facing the greatest difficulty will be the long tail of smaller vendors selling niche tools for non-critical business workflows. Customers may find it increasingly difficult to justify paying for those products when a growing number of cheaper, integrated, or AI-generated alternatives become available.
