Competitive Strategy in the Era of Applied AI

AI has entered a phase of deeper interplay with the rest of the economy, with huge profit zones at stake; but AI is more likely to destroy an old competitive moat than provide a new one. This poses fundamental questions for the competitive strategy playbook.

The recent emergence of powerful AI generative, reasoning, coding, and orchestration capabilities has fundamental implications for competitive strategy for businesses of all stages and industries. Below, we consider those implications, with examples pointing to three frameworks: Michael Porter’s Five Forces, Peter Thiel’s Zero to One, and Adrian Slywotzky and David Morrison’s The Profit Zone.

By compressing development cycles, AI upends long-valued pillars of differentiation and threatens to narrow the competitive strategy playbook. Ironically, as AI capabilities improve and become more widely available, they no longer differentiate. Sources of durable competitive advantage are becoming much scarcer. We conclude that a new class of architectures and systems will become structural keystones with the potential to drive large profit zones in the coming era, and that patent rights can provide a durable moat around them.

Overview of Frameworks

Competitive strategy frameworks provide a structured way to assess how a company creates, captures, and defends value relative to its competitors, helping leaders make better strategic decisions and investors identify businesses with durable competitive advantages and superior long-term returns. We draw from three different frameworks to examine a common question: Where can durable strategic control reside as AI makes capabilities easier to replicate? We first explain the frameworks, then examine how AI weakens traditional sources of differentiation, and finally distinguish enduring architectures and systems from the patent rights that can protect them.

Michael Porter’s Five Forces argues that a company’s profitability is shaped not simply by its own capabilities or its direct competitors, but by the structure of the industry in which it operates. Five forces determine the intensity of competition and the ability to capture economic value: rivalry among existing competitors, the threat of new entrants, the threat of substitute products or services, the bargaining power of buyers, and the bargaining power of suppliers. A strong strategy positions a company to defend against these forces or reshape them in its favor, creating a sustainable competitive advantage. The central lesson is that understanding and influencing industry structure is as important as competing effectively within it. The framework, a classic in MBA curricula, is quite useful in analyzing competition for mature businesses operating within well-established industries.

Peter Thiel, who has been referred to as the “Prophet” and the “Philosopher King” for his record as a visionary investor in Silicon Valley, argues that the greatest businesses are built not by competing within established markets, but by creating something genuinely new. In Zero to One, referred to by some of the most successful company founders in history as their “Bible,” Thiel distinguishes “zero to one” innovation—creating new technologies, markets, or capabilities—from “one to N” progress, which replicates or scales what already exists. He contends that enduring value comes from building a differentiated business with the characteristics of a durable monopoly, supported by advantages such as proprietary technology, network effects, economies of scale, and/or a powerful brand. A central lesson is that the greatest strategic advantage is not simply winning the competition, but creating something so distinctive that competition becomes far less relevant. As AI is poised to unlock entirely new solutions to persistent problems, Thiel's framework encourages entrepreneurs to ask seven questions:

The Engineering Question Can you create breakthrough technology?
The Timing Question Is now the right time?
The Monopoly Question Can you start with a big share of a small market?
The People Question Do you have the right team?
The Distribution Question Do you have a way to sell the product?
The Durability Question Will your position still be defensible in 10–20 years?
The Secret Question Have you identified an opportunity others don't see?

Adrian Slywotzky and David Morrison’s The Profit Zone: How Strategic Business Design Will Lead You to Tomorrow’s Profits occupies a middle ground, receiving widespread praise for helping leaders of mid-sized companies recognize and respond to inevitable shifts in where profits can be captured. The authors argue that sustained profitability depends less on maximizing market share than on positioning a business where it can create and capture the greatest economic value. Companies must understand what customers truly value, identify the most attractive profit models, and establish strategic control points that protect their ability to earn superior returns. A control point is not merely something a company does well. It is an asset or position that allows the company to retain economic value because other participants cannot easily replicate or bypass it.

Taken together, these frameworks help leaders at all stages, from seed stage startups to mature companies in large industries, think more rigorously about competitive strategy. There is one thing they have in common: the holy grail is to have sources of durable strategic control over profitable offerings.

AI Wreaks Havoc on Strategic Control

AI destroys moats. It is far more likely to damage an existing moat than provide a new one. This is the challenge for all companies in the coming era, when we can expect much deeper interplay between AI and the rest of the economy. AI-enabled development has weakened technology, product, software, speed, team, operational excellence, and even intelligence itself as sources of differentiation. Company leaders should fear rapid erosion of profit zones. Investors should fear widespread commoditization.

AI challenges traditional differentiation in three connected ways: It compresses the creation of technology, software, and hardware; weakens platform and distribution advantages; and changes the economics of talent, intelligence, and data. We examine each in turn, using the competitive strategy frameworks to explain the implications. Speed connects the first group: faster development makes capabilities easier to match and harder to defend.

Technological Breakthroughs: Know-how and Trade Secrets

Traditionally, technological breakthroughs required unusual expertise, large engineering or scientific teams, specialized infrastructure, and years of R&D. AI now delivers breakthroughs where the most brilliant humans struggled for decades:

DeepMind’s AlphaFold cracked the longstanding protein-folding prediction challenge, predicting the structures of hundreds of millions of proteins and accelerating drug discovery and biological research.
An internal OpenAI model recently proposed a solution to the Navier-Stokes Millenium Problem, shaking the mathematics community. AlphaProof and AlphaGeometry have solved extremely difficult math problems and demonstrated exceptional capability for automated reasoning.
AI has helped theoretical physicists discover new solutions to complex equations and identify previously unknown patterns in physical systems.

AI is capable of uncovering solutions, patterns, and possibilities that have long eluded humans. For those seeking to answer Thiel’s engineering question in the affirmative, a breakthrough is therefore only the starting point. If exceptional reasoning capabilities can be applied to digitized public knowledge, an important strategic question follows: What prevents competitors from discovering or reproducing the same capability? The durability of a secret depends on what others can access and replicate, not simply on the difficulty of the original discovery.

Software

As most of the world’s assets and workflows are now software-defined, software has been a key source of competitive advantage for many of the most successful businesses of the last two decades. SaaS businesses have formed a big component of many private equity portfolios. Consulting businesses have profitably leveraged offshore teams of thousands of programmers. Companies in many industries have sought to differentiate products, services, or platforms through intelligence and connectivity features, enabled by software.

Generative AI reduces the cost of producing software dramatically, even at the scale of entire platforms. A single sophisticated developer can now launch a large team of coding agents that generate and debug code; prototype products rapidly; build interfaces; analyze large quantities of information; automate testing and documentation; and interact with APIs and other software systems.

Agentic coding has also unlocked exponential improvements in data and systems integration. This could have profound implications in industries like health care and banking, where integration complexity among a myriad of siloed workflows has long served as a de facto moat. A talented programmer can now deliver users what previously required a large organization.

To sum up, agentic coding has the potential to unlock huge performance gains for adopters, but those gains are immediately available to any talented developer who has access to insight about how an industry operates.

Hardware

Hardware design cycles are also compressing as AI moves from assisting with documentation and analysis to actively generating, testing, and optimizing physical designs. As a result, AI is compressing the iteration loop across the entire physical product-development process, with varying effects across categories as illustrated in the table below. As coding, configuration, simulation, optimization and verification become increasingly automated, engineers can evaluate more alternatives before committing to expensive fabrication. That enables faster product launches, lower development costs, more frequent hardware revisions and increasingly customized physical products.

Hardware	How AI changes the cycle	Quantitative signal / caveat
Chips / SoCs	Assists RTL coding, floorplanning, timing optimization and debugging.	Up to 10× acceleration reported for certain chip-design tasks—not total chip development.
PCBs	Automates component placement, routing and constraint checks.	Cadence reports up to 2× faster time to market and 15× productivity for its agentic PCB/packaging platform.
Mechanical CAD	Generates geometry alternatives and optimizes weight, strength and cost.	Some vendor-reported workflows reduce selected tasks from days to under an hour; not a universal end-to-end metric.
FPGA / digital logic	Generates HDL code and tests; helps diagnose and correct verification failures.	Research prototypes show automated design-and-verification workflows; no broadly comparable cycle-time figure.
Analog / mixed-signal	Explores circuit configurations and assists layout optimization under electrical constraints.	Research reports 2–3× faster layout iteration on selected tasks; results vary by design.
Optics / antennas	Optimizes geometry and uses surrogate models to reduce expensive simulation runs.	Potentially much faster design-space exploration; no single reliable cross-industry multiplier.
Robotics / electromechanical	Coordinates mechanical, electrical and control design; tests in simulation and digital twins.	More parallel iterations and earlier integration; physical and safety validation remain necessary.
Industrial machinery	Reuses engineering knowledge and automates configurable designs and customer-specific changes.	Faster customization is an emerging benefit; robust, comparable cycle-time figures are not yet standardized.

We expect the next major step will be agentic orchestration across the full stack—from requirements and chip architecture through PCB layout, mechanical design, embedded software, simulation and manufacturing—so that changes in one subsystem can automatically trigger coordinated changes in others. Physical production, material constraints and certification will still impose limits, but the engineering work leading to those milestones can increasingly run at machine speed. Because hardware is designed in software, new hardware also needs a durable competitive moat beyond unlocking improved performance or new capability.

Speed

As the minimum resources needed to make a breakthrough, build a serious product, or create new software or hardware have plummeted, timing at all stages of R&D is also compressed. AI dramatically compresses R&D and overall market timing cycles. Previously, a company might have had years to exploit a technological discontinuity before competitors caught up. With foundation models, APIs, open-source models, AI coding agents, hardware design tools, and computational infrastructure widely available, the time from new capability to commercial application to competitive imitation can be extremely short. Product monopolies have become almost impossible to maintain, if they are based on capabilities alone. Speed is necessary to maintain parity with accelerated competitors, but it is insufficient to differentiate.

Compressed development cycles have huge strategic implications:

Four of Porter’s Five Forces become much stronger: new substitutes are far easier to make, competitors can rapidly reach parity without large resource drains, buyers have more bargaining power because they can build their own capabilities, and the threat of new entrants goes off the charts.
The bar is raised dramatically for Thiel’s engineering, timing, people, and durability questions. Every move, no matter how fast, has a faster follower than ever before.
In Slywotzky and Morrison’s framework, strategic control points protect a company’s ability to capture value within profit zones. Many of the most attractive of those control points are gone.

This could upend some of the longstanding pillars of the competitive strategy playbook.

Natural Monopolies and Platforms: Network Effects and Switching Costs

If product monopolies become extremely difficult to maintain, businesses will need other control points. Over the last 30 years, some of the most successful businesses in history have established powerful platform positions strongly supported by network effects, where participation becomes more valuable as other users join, and high switching costs, where users face effort or losses when moving to another services. These have been two of the most durable sources of competitive advantage in the digital economy. They are so powerful that a surprisingly large fraction of the value of the entire economy has concentrated in a small number of platform companies positioned at distribution bottlenecks, such as in search, e-commerce, sales and marketing, and media.

AI weakens both network effects and switching costs. By rapidly distilling the accumulated content of the Internet into general-purpose models, AI makes knowledge, functionality, and capabilities that once resided behind individual platforms’ walls increasingly accessible outside those platforms. Users can obtain answers without visiting the original sites, replicate favored features without adopting the underlying software, and move between competing services with far less friction. As a result, content accumulation and feature breadth alone are becoming less defensible advantages.

The question is no longer only whether a business can establish a monopoly position, but whether that position can endure. This goes to the heart of Thiel’s durability question and removes another large swath of strategic control points that could protect profit zones.

The next question is where strategic control will reside if existing platform advantages weaken. The ability to orchestrate assets, workflows, proprietary data, and real-world systems into integrated solutions opens the possibility for entirely new categories of software and media. These could repeat the pattern of concentrated platform power, producing even stronger monopolies, or restore a better balance among creators, distribution providers, and users.

Some of today’s platforms are quite unpopular, and small teams of motivated developers are already working with the capabilities of AI to replace them. If successful, those efforts, along with agentic purchasing capabilities (discussed below), will increase the bargaining power of buyers in the Five Forces framework. For the first time in decades, even the giant natural monopolies are subject to greater competitive forces.

Distribution

For the last three decades, distribution has been a reliable source of strategic control. In the modern economy, distribution is strongly associated with the natural monopolies that created extreme distribution bottlenecks in some categories. Value concentrated in search, e-commerce, content, and sales and marketing. If the effects that created those monopolies diminish as contemplated above, where does that leave Thiel’s distribution question?

As to distribution more generally, AI can generate personalized outbound sales, marketing content, advertising variants, product demonstrations, customer-support responses, onboarding materials, sales proposals, lead qualification, and market research. A tiny company can potentially perform activities that previously required a large sales and marketing organization.

Notably, while AI lowers the cost of distribution, it doesn’t necessarily create distribution. Big questions remain about how distribution will play out in a world where agentic AI takes over more of the work of discovery, procurement, negotiation, and purchasing for buyers. Agents working effectively on behalf of buyers will increase their bargaining power, one of the Five Forces.

Distribution is also often tightly associated with a powerful brand, a strategic control point emphasized by Thiel and in the Profit Zone framework. If agents do much of the purchasing, will they care about brands or relationships, or will they rapidly switch to the least-cost provider of the minimum necessary capability? At best, distribution remains a much more uncertain source of long-term competitive advantage than before.

Talent and Organization

Thiel’s people question and the Profit Zone framework emphasize aligning the strengths of a team with a potentially profitable market position. Generative AI fundamentally changes the economics of teams because one person can now perform work that previously required many specialists. A sophisticated founder can use AI to act partially as programmer, researcher, analyst, designer, copywriter, customer-support agent, QA engineer, data analyst, and project manager. That diminishes the advantage of having a large organization.

An important counterpoint is that AI increases the value of exceptional people. If everyone has access to the same AI, the scarce resource becomes the person with great judgment, creativity, vision, domain knowledge, ability to identify valuable problems, ability to coordinate humans and AI, and ability to make decisions under uncertainty. The talent question becomes: “Do you have people capable of directing extremely powerful tools toward something economically valuable?"

We are long believers in inventive talent. As the path from invention to a working system compresses, the value of the invention itself increases, provided that a durable moat can be wrapped around it.

Intelligence and Data

Proprietary intelligence is a longstanding strategic control point for many businesses. Ironically, AI may simultaneously help a company improve its intelligence and diminish its value. Imagine building a company or product because today's models cannot reliably perform a particular task. Shortly afterwards, model capabilities improve and replace the entire value proposition. Companies using AI simultaneously benefit from technological progress and are threatened by it.

An extension of this principle is Satya Nadella’s “reverse information paradox,” which captures a central tension of the AI era: as information becomes abundant, its economic value shifts elsewhere. The core of the paradox is that the act of using AI creates a risk that a company’s own proprietary knowledge, operating processes, and insights will be absorbed, distilled, and/or replicated by the very AI systems they use. What once served as a durable intelligence advantage for the company now diffuses first into the general models (which all distill from each other) and then into the hands of the competitors and customers. In this scenario, instead of providing a strategic control point, intelligence actually strengthens competitive forces. Every company faces a prisoner’s dilemma. It must adopt AI to maintain parity with its cohorts, but in doing so it risks losing control of its own intelligence.

Nadella proposed creating a “trust boundary” around AI models, and various industry players have proposed security and governance frameworks designed to protect proprietary data. The problem is that distillation of intelligence isn’t a security flaw, it’s inherent to what an AI model does.

When much of the information of the world can be accessed by AI, the mere aggregation of data becomes a weaker differentiator; however, certain kinds of data remain promising sources of more durable value. We believe real time contextual data about assets and workflows will become highly valuable, because it is very difficult to replicate in AI models, and it unlocks new, closed-loop agentic architectures where context-capable AI responds to situational changes and instantly acts upon software-defined assets and workflows. Real time data is part of a new kind of agentic architecture that is poised to reshape many industries. As with the architectures discussed below, the enduring usefulness of these inputs and a company’s ability to maintain exclusive control over them are separate questions.

More on Durability

Whether the starting point is industry structure, strategic control points, or company differentiators, the analysis converges on Thiel’s durability question. AI potentially undermines technological knowledge, speed, differentiated products, software, platform and network effects, distribution advantages, large teams, and intelligence itself as pathways to durable control. The remaining question has two parts: Which capabilities will remain economically important, and what will prevent competitors from capturing the same value? The following sections examine AI models, enduring architectures and systems, and the patent rights that can protect them.

AI Models Are Interchangeable

The effects of AI are applicable to AI models and infrastructure. As Jensen Huang observed at Davos in early 2026, the cognitive and agentic capabilities of AI are delivered by a converged set of technologies in four increasingly standardized layers: energy, chips, cloud/networking, and models. Despite huge capability gains, the models themselves are still relatively undifferentiated: Nearly identical interfaces provide access to very similar capabilities. In the Five Forces framework, model providers face fierce competition from other providers; new entrants can rapidly acquire comparable capabilities; buyers have leverage because they can create their own substitutes; and a powerful upstream supplier still has pricing power over the chips needed to keep improving.

In other words, there are large doubts about whether an AI model itself can serve as a durable strategic control point. As the interplay of AI with the rest of the economy increases, models are likely to serve as interchangeable components of larger systems, commanding modest profits.

Durable Architectures and Systems

As noted above, we believe one durable feature of the coming era of Applied AI will be a new class of architectures where context capable AI responds to situational changes and instantly acts upon software-defined assets and workflows. We expect that more and more new systems will be constructed upon these architectures, unlocking large performance gains, as is already happening in some fields.

Those architectures and systems will reflect widespread technology convergence and the deep interplay of AI with systems and workflows of the rest of the economy. With variants, certain complementary capabilities, such as sensor fusion that provides real time inputs to AI; digital twins that allow real time monitoring, planning and simulation; and systems that intertwine governance with AI are likely to endure as important capabilities.

The highest-level systems will link and orchestrate today’s siloed systems to optimize across them, unlocking large performance gains. Eventually, many of today’s systems will likely integrate into new, higher-level systems, such as control towers that unify enterprise or market functions.

An architecture can endure without giving any one company a durable competitive advantage. We expect these architectures to remain important, but they are widely available, and adopting them may be necessary to survive in the era of Applied AI without preventing competitors from doing the same. Their economic importance and their defensibility are therefore separate questions. Patent rights provide the potential source of exclusivity examined next.

Patents Provide Durable Moats for Architectures and Systems

The patent system provides a special kind of durable moat: a legal monopoly that applies for up to 20 years, regardless of how or by whom a covered method or system is made. A patent moat is only available to the first mover (not an early mover, or a fast mover—the very first one) who also takes the initiative to work through the complex patent process. As a result, patent moats are inherently scarce.

For a patent moat to have durable value, it must cover something that itself endures as an important part of an economic system. The vast majority of patents relate to incremental product features or technology improvements. This is why players in venture ecosystems that focus on product-market fit only rarely, if ever, experience the enduring power of a true patent moat. For product-oriented patents, the compressed development cycles described above will probably render incremental capabilities and the corresponding patents less valuable than in the past.

On the other hand, patents on enduring, systemic capabilities, like the new agentic architectures and systems described in the previous section, provide a powerful, long-lasting competitive moat around those capabilities. The distinction is not simply between product patents and architectural patents: Durable value depends on protecting capabilities that remain economically important, as described above.

Those who hold architectural and system patents can enjoy enduring differentiation of their offerings (higher margins, market share, and multiples); negotiating leverage in how platforms are organized and ecosystems are structured; and the ability to command participation in revenue streams outside their own roadmaps.

We believe a new class of agentic architectures that underpin Applied AI, and the systems built upon them, will drive a massive new profit zone in the coming years. The strategic opportunity is not simply to adopt these architectures, but to secure durable control over economically important capabilities within them. Architectural and system patents can provide that moat. This brings the three frameworks back together: shaping industry structure, building a differentiated position that can endure, and protecting the control points where profits accrue.

For leaders and investors, the practical test is to separate three questions: What performance gain does AI enable? What prevents competitors from delivering the same outcome? And who controls the capabilities needed to capture the value? Adopting AI addresses the first question. Durable competitive strategy requires answering the other two.