Energy has always been the foundation of civilization. From fire to fossil fuels to electricity, each major leap in energy systems has unlocked new phases of human progress.

But today, energy is no longer just enabling growth, it is becoming the primary constraint on it.

We are entering a moment where three powerful forces are converging:

• Geopolitical instability, including rising tensions in the Middle East and potential disruptions to global energy supply

• The rapid acceleration of artificial intelligence, driving unprecedented demand for electricity

• Aging infrastructure that was never designed for this level of scale, speed, or complexity

In prior eras, energy innovation was about efficiency and cost. Today, it is about capacity, resilience, and control.

The rise of AI and hyperscale data centers has transformed energy from a background utility into a strategic asset. Compute is no longer limited by chips alone. It is constrained by power, cooling, and grid access.

This shift marks the beginning of a new phase in the evolution of energy systems, one where energy and intelligence are deeply intertwined. We are not running out of compute. We are running into the limits of energy.

The Historical Significance of Energy

In early civilizations, humans primarily focused on survival, constantly struggling to find a stable food supply. The advent of agriculture transformed this dynamic, allowing populations to grow from small, nomadic groups to vast, settled communities. For example, the global population increased from about 5 million in 10,000 BCE to over 250 million by the first century CE, largely due to increased food production from agricultural advancements. During this period, humans relied on animals for labor, using their strength to plow fields and transport goods. Energy consumption was limited by the availability of food for both humans and animals, making food the primary energy source.

The Industrial Revolution marked a pivotal transformation, expanding human energy use beyond biological sources to include fossil fuels such as coal and oil. This era introduced groundbreaking technologies like steam engines and automobiles, revolutionizing transportation and manufacturing. These advancements catalyzed a period of remarkable growth and innovation, fundamentally changing economic structures and daily life, and setting the stage for the modern industrialized world. This shift not only increased the scale of human activity but also began the complex relationship between energy consumption and environmental impact, framing many of the challenges and opportunities we face today.

The Modern Energy Landscape

The modern energy landscape is being reshaped by a surge in demand driven by population growth, electrification, and rapid technological advancement. From electric vehicles to cloud computing and artificial intelligence, new systems are increasingly energy-intensive. Despite significant expansion in clean energy, supply is struggling to keep pace with this accelerating demand, leading to visible strain on power systems globally.

By 2030, data centers in the United States are projected to consume approximately 8% to 10% of total electricity demand, up from roughly 3% in 2022. This sharp increase is driven primarily by artificial intelligence and other compute-intensive workloads. AI is emerging as the dominant source of incremental demand, with data centers supporting these workloads expected to more than double their electricity consumption over the coming decade due to the higher energy density required for training and inference at scale.

Historical infrastructure cycles highlight the magnitude of this shift. Early internet infrastructure operated at relatively modest scale, requiring only a few gigawatts of power. The rise of cloud computing expanded this to tens of gigawatts over more than a decade. Today, we are entering a new phase defined by AI, sometimes referred to as Cloud 2.0 or Internet 3.0, where demand is accelerating much faster. Instead of multi-decade buildouts, infrastructure is scaling within just a few years, with projected demand reaching tens to potentially hundreds of gigawatts globally.

Balancing the Trade-offs in Energy Production

Despite significant investments in renewable energy sources such as hydro, solar, and wind, these technologies have not yet supplanted traditional fossil fuels as the dominant energy sources. As of 2024, the global energy mix remains predominantly reliant on fossil fuels. Coal, oil, and natural gas together account for approximately 79% to 80% of total primary energy consumption, spanning electricity, heating, transportation, and industrial use. Despite rapid growth in renewable energy, this share has only declined modestly over the past three decades, highlighting the structural inertia of global energy systems.

The real-time operation of the electricity power grid presents a challenge for aligning clean energy production with daily consumption patterns. Solar energy is only available during the day, wind energy typically increases at night, and hydroelectric power fluctuates seasonally, leading to limitations and unreliability in their production timing. Energy storage systems, such as batteries, play a crucial role in addressing this mismatch by storing excess energy during low-demand periods and releasing it during peak times. However, utility-scale batteries typically have a round-trip efficiency of about 80%, meaning that 20% of the stored energy can be lost in the process of charging and discharging.

To ensure sustained energy growth while fostering a sustainable environment for future generations, it is crucial to balance and continually enhance our mix of energy sources by leveraging technology to optimize energy sources and consumption.

The Convergence of Energy and Intelligence

As energy systems face increasing strain, a new layer of innovation is emerging at the intersection of energy and computation. Artificial intelligence is not only driving demand for energy. It is also becoming a critical tool for managing it.

Energy-Integrated Infrastructure

Data centers are beginning to co-locate directly with power generation. Instead of relying solely on the grid, they are being built alongside:

• Renewable energy farms paired with storage

• Hydroelectric facilities

• Next-generation nuclear systems

This reduces transmission bottlenecks and improves reliability at scale.

Intelligent Grids and Software-Defined Energy

The integration of software into energy systems is accelerating.

AI-driven platforms can optimize energy flows in real time, improving efficiency across generation, transmission, and consumption. This transforms the grid from a rigid system into a programmable one, where intelligence plays a central role in maintaining stability.

Energy-Aware Compute and Model Efficiency

As energy becomes a constraint, efficiency becomes a competitive advantage.

We are seeing increased focus on:

• More efficient model architectures

• Hardware optimization for lower energy consumption

• Workload scheduling based on energy pricing and availability

The next generation of AI systems will not just be more powerful. They will be more energy efficient by design.

New Bottlenecks, New Opportunities

As the system evolves, value is shifting to previously overlooked components:

• Cooling technologies, including liquid and immersion cooling

• Power electronics and transformers

• Long-duration energy storage solutions

These are no longer peripheral technologies. They are becoming essential to scaling both energy and compute infrastructure.

Energy: The Foundation of the Visionary Economy

Energy has always shaped the trajectory of human progress. But in this new era, its role is expanding once again.

We are moving from a world where energy powers industry to one where it powers intelligence.

At the same time, geopolitical uncertainty, infrastructure constraints, and accelerating demand are redefining the limits of growth. The challenge is no longer simply generating more energy. It is orchestrating a system that can deliver it reliably, efficiently, and at the scale required by modern computation.

In the Visionary Economy, energy is no longer a background input. It is a foundational layer of innovation.

The most valuable systems of the future will not be defined solely by better algorithms or faster chips. They will be defined by the integration of:

• Compute

• Energy

• Infrastructure

• Intelligence

The winners will be those who understand this convergence and build systems that align multiple layers seamlessly. Because in the age of AI, energy is no longer just part of the economy. It is a core component that defines what is possible.

For decades, elite services like legal counsel, accounting, consulting, insurance and financial advisory have been powered by human labor. Expertise was scarce. Time was billable. Access was expensive.

That model is breaking.

We are entering the era where AI does not just support services. It delivers them.

The Great Flip: From 80% Human to 80% AI

Traditional service firms run on an 80 percent human model. Junior associates grind through research. Analysts reconcile data. Consultants build decks. Partners review and advise.

Technology has been a tool, not the engine.

AI is changing the ratio. We start to see AI native service firm that runs 80 percent on software and 20 percent on human oversight. Humans design frameworks, handle edge cases, build relationships, and make judgment calls. AI performs the repeatable cognitive labor at scale.

• Contract review

• Audit testing

• Due diligence

• Financial modeling

• Regulatory compliance checks

• Customer support triage

These become software functions.

The cost structure collapses. The margin profile expands. Accessibility explodes.

This is not incremental efficiency. It is structural deflation in high end services.

From “Software Is Eating the World” to “AI Is Eating Services”

When Marc Andreessen wrote that software is eating the world, he described how code replaced physical distribution and manual processes.

Software scaled products. AI scales expertise. That is a bigger shift.

SaaS made tools subscription based and continuously improving. AI makes judgment, analysis, and reasoning subscription based and continuously improving.

We are watching the birth of Expertise as a Service.

Subscribe now

Share

The Venture Lens: Where the Value Will Accrue

From a VC perspective, this transformation creates three massive opportunity zones.

1. AI Native Service Platforms

Startups that rebuild legal, accounting, compliance, insurance and advisory services from first principles.

Not a traditional firm with AI layered on top. A software company that happens to deliver services.

These companies will show:

• High gross margins

• Usage based pricing

• Continuous model improvement

• Embedded data flywheels

This is the extension of the traditional SaaS subscription model except it does not just sell software, it sells solutions directly. The winners will not sell subscription or hours. They will sell outcomes.

2. AI Infrastructure for Service Firms

Not every law firm or accounting firm will disappear. Many will modernize. There is a massive opportunity in building a new AI infrastructure layer that powers them:

• Secure model orchestration

• Domain specific fine tuning

• Audit trails and explainability

• Workflow automation

• Human in the loop controls

Enterprise AI infrastructure will become the backbone of professional services.

This is where durable venture scale businesses can emerge.

3. Hybrid AI First Firms

A new category will rise. Firms that look traditional externally but are structurally AI powered internally.

High Touch + High Tech

On the surface:

• Partners

• Client relationships

• Advisory retainers

• … similar to a traditional service firm

Under the hood:

• AI driven research

• Automated compliance

• Model generated reporting

• Decision support systems

This architecture fundamentally reshapes the economics.

• Lower cost structure.

• Higher operating leverage.

• Software like margins.

The result is disruptive pricing without sacrificing quality, and in many cases, improving it.

These firms will not compete on hours billed. They will compete on intelligence delivered.

Recommendations for Founders

If you are building in this space, focus on:

• Start with workflow, not models - Clients buy solutions, not algorithms

• Embed deeply into customer systems - APIs and integrations are defensibility

• Own the data layer - Data improves models. Models improve outcomes. Outcomes generate more data

• Design for human trust - Explainability, auditability, and compliance are non negotiable

• Price for value delivered - Do not underprice AI. If you reduce legal spend by 60 percent, capture a share of that delta

The moat is not just technology. It is trust plus integration plus domain depth.

The Economic Implication

Small businesses that could never afford top tier legal advice will now subscribe to it.

Startups will automate compliance instead of hiring teams.

Enterprises will compress layers of middle management.

This is productivity expansion at scale.

It also reshapes labor markets. High end knowledge workers will move up the stack toward strategy, creativity, relationship building, and oversight.

Routine task work will decline in value.

The Visionary Economy rewards those who orchestrate intelligence, not just perform tasks.

Conclusion: AI Defined Services

The move from 80 percent human to 80 percent AI will not happen overnight. Regulation, trust, and inertia slow transitions.

But the direction is clear.

High-cost services become software like…
Expertise becomes programmable.
Advisory becomes scalable.

In the Visionary Economy, value accrues to those who redesign industries around intelligence.

The question is not whether AI will transform services. The question is who captures the upside. And from a venture lens, the opportunity is just beginning.

What if the education system we’ve trusted for centuries is no longer sufficient for the future we’re entering? Imagine an education model designed like a factory, primarily focused on churning out workers trained to perform specific tasks within a narrow set of roles. This system, founded on a one-size-fits-all approach, including standardized tests and predetermined career paths, has been the backbone of our society. But now, as AI and automation become ever more integral to our economic fabric, we find this model drastically inadequate. Now, we are on the brink of a fundamental shift, we need to reevaluate both the purpose and methods of education in an economy driven by innovation knows no bounds.

The Rapid Shift in Career Paradigms

Our educational methodologies have become stagnant, entrenched in a cycle of standardized learning and mere replication of existing knowledge. Students are shaped to fit a pre-existing mold, reduced to cogs in an industrial machine, crafted to function as repetitive production engines in society.

In the traditional career model, we envision a linear path: graduate from university by around age 22 and embark on a roughly 40-year career, climbing a vertical ladder from junior to senior positions within the same field. This approach made sense in an era where societal advancements were gradual, and careers were designed around mastering and performing a specific job function over a lifetime. However, in today’s world, where technological advancements occur at warp speed, this static career model and implementer mind set are increasingly obsolete.

The speed of technological adoption has surged dramatically over the past two centuries, profoundly altering our daily lives and the very fabric of society. Imagine the early telephone, a groundbreaking invention of the 19th century, took a staggering 75 years to connect 100 million users. Fast forward to the late 20th century, when the personal computer (PC) burst onto the scene, revolutionizing how we gather, process, and share information. This digital marvel reached 100 million users in just 15 years. As we stepped into the 1990s, the internet began weaving its way into the public sphere, swiftly becoming a cornerstone of global communication, knowledge sharing, and commerce. Astonishingly, it crossed the 100 million user mark within about 7 years. Today, the pace of adoption has reached breakneck speeds with artificial intelligence leading the charge. ChatGPT, a cutting-edge AI, shattered previous records by amassing 100 million users in a mere 2 months.

Traditional job roles, once seen as lifelong vocations, are now fleeting. The idea of a “job for life” is becoming a relic of the past. As technology leaps forward, it outpaces the slow-changing social structures that once supported stable, long-term employment. Today, we are entering an era where flexibility and adaptability are paramount.

Careers now require individuals to pivot and evolve continuously, acquiring new skills and adapting to varied roles and industries. This shift is moving us away from being mere implementers of tasks to becoming lifelong learners and versatile contributors in a rapidly changing world. The reality is stark: adapt quickly or risk obsolescence. As we navigate this dynamic employment landscape, the challenge and opportunity lies in our ability to keep pace with the relentless speed of change, ensuring we remain relevant and resilient in the face of constant technological innovation.

Redefining in the AI Era

In the dance of progress, where do humans and AI partner to perform their best? Unlike machines, which excel in storing vast amounts of data and executing repetitive tasks flawlessly, humans bring to the table an “abstraction” ability that enables us to synthesize diverse information and solve problems creatively. Henry Ford, a pioneer of modern industry, once famously said that he didn’t need to know everything himself; with a simple press of a button, he could summon an expert to provide answers to any of his questions. Today, AI steps into those shoes, serving as our modern-day ‘button-pressed’ experts, thus freeing us to focus on innovation and strategic thinking.

In this rapidly evolving landscape, our educational system needs transformations. It shouldn’t aim to churn out more efficient human implementers. Instead, it should nurture learners’ ability to think creatively and use knowledge innovatively. Our children need to be taught not just to answer questions, but to question the answers, and equally importantly, to master the art of asking the right questions. After all, framing the right question is often half the battle won.

Traditional educational assessments too often prioritize memorization, undermining the real essence of learning: comprehension and application. In the era of AI, where factual knowledge is at everyone’s fingertips, the true value lies in making connections, thinking outside the box, and navigating complex challenges. This shift from mere information regurgitation to fostering deep understanding and innovative application is crucial. By redefining our approach to learning in this way, we prepare ourselves not just to coexist with AI but to collaborate effectively, ensuring that both humans and artificial intelligence can thrive synergistically in a future crafted by both.

As we reconsider what we value in education, it’s clear that the focus should shift from preparing students for specific jobs to equipping them with the skills necessary to thrive in an AI-enhanced world. This includes fostering design thinking, creative problem-solving, and an entrepreneurial mindset. Moreover, the integration of AI into education can provide personalized learning paths that adapt to individual needs, further enhancing engagement and effectiveness.

The essence of humanity lies in our innate capacity to evolve — to become, rather than merely to be. We are more than just “human beings”; we are “human becomings,” perpetually striving to improve and transform into better versions of ourselves.

Investing in Human Capital: An AI-Era Education Strategy

From a venture perspective, education is not about curriculum reform. It is about capital allocation. The question is simple: what kind of human capital will compound in an AI-driven economy?

1. Design Thinking Over Task Training

As AI absorbs routine execution, the highest-return skill shifts to problem framing, systems thinking, and creative synthesis. Education should train students to identify opportunities, not just complete assignments. In venture terms, we are investing in founders of ideas, not operators of instructions.

2. AI as Leverage, Not a Shortcut

Personalized learning powered by AI is not about convenience. It is about increasing cognitive ROI. When students can iterate faster, simulate scenarios, and explore ideas dynamically, learning becomes compounding rather than linear. AI becomes a force multiplier for curiosity.

3. 0→1 Capability

In venture capital, outsized returns come from nonlinear breakthroughs. Education should cultivate the ability to build from zero to one, not just optimize from one to ten. That means encouraging experimentation, tolerating failure, and rewarding original thinking over standardized correctness.

4. Lifelong Learning as Career Infrastructure

Careers are no longer vertical ladders. They are dynamic portfolios of skills. Education should become a continuous platform rather than a one-time phase. The most valuable individuals will not be those with static expertise, but those who can repeatedly re-skill and reposition themselves as markets evolve.

5. Global and Cross-Disciplinary Fluency

The most successful companies operate across borders and industries. Human capital should do the same. Cultural agility, interdisciplinary thinking, and collaborative intelligence will define leadership in a distributed, AI-enabled world.

As AI reshapes the economic landscape, education cannot remain static. It needs to evolve from training for tasks to cultivating judgment, creativity, and resilience. The goal is not to prepare students to compete with machines, but to guide them. In the Visionary Economy, the future belongs to those who can think deeply, adapt continuously, and choose wisely in a world of accelerating possibility.

The Internet era reshaped how the world connects. The internet democratized information, outsourced coordination, and linked global markets at unprecedented scale. It enabled people to communicate, transact, and collaborate across borders with speed that once seemed impossible. Yet throughout that transformation, innovation remained fundamentally human-led. Technology amplified human effort, extended human reach, and accelerated human productivity. It did not replace the human mind at the center of value creation.

Today, we are entering a fundamentally different phase.

The AI economy is not a continuation of the internet story. It introduces an entirely new dimension of optimization by adding AI as a parallel intelligence and execution layer alongside humans. AI does not simply help people work faster. It performs cognitive work itself. Where the internet connected people, AI embeds intelligence directly into systems. Where the dot-com era focused on access and efficiency, the AI era reshapes execution, decision-making, and the structure of work itself. This is not an upgrade in connectivity. It is a transformation in how value is created and who or what creates it.

The Workforce Transformation: Humans, AI, and Robotics

For the first time in modern economic history, the workforce is expanding beyond humans to include AI agents and robotics as active contributors to production and decision-making. This shift fundamentally changes how organizations allocate resources and think about cost structures.

Historically, technology spending accounted for a relatively small share of company revenue, often in the range of single-digit percentages, while human labor represented the largest expense, frequently 30 to 40 percent or more in knowledge-driven industries. AI disrupts this long-standing balance by positioning itself as a form of virtual labor, one that increasingly performs tasks once reserved for human workers.

As AI agents and autonomous systems take on cognitive and operational roles, businesses begin to evaluate technology investment not as overhead, but as an alternative to traditional workforce costs. This marks a structural shift in how value is created, how budgets are planned, and how organizations define productivity in the AI economy.

Consider this scenario: a CEO evaluating resource allocation might now debate between investing in an AI Agent or hiring additional customer service representatives. AI agents, equipped with natural language processing, can manage customer inquiries at scale, operate 24/7, and continuously improve through machine learning. Similarly, marketing departments might replace entry-level analysts with AI tools capable of running sophisticated campaigns, analyzing market data, and optimizing strategies in real-time.

This shift marks a structural change in how budgets are planned, how productivity is measured, and how organizations define growth. The internet era disrupted industries by expanding access. The AI economy disrupts by automating and transforming execution itself. Given the scale of human-related costs in most organizations, the economic incentive to adopt AI is not speculative. It is structural.

Beyond Automation: A Cross-Industry Transformation

The impact of the AI economy extends far beyond any single function or sector. In healthcare, AI-driven diagnostics and treatment planning systems accelerate decision-making and reduce dependence on manual analysis. In manufacturing, intelligent robotics optimize production lines, operate continuously, and adapt dynamically to changing conditions. In finance, AI systems manage risk, detect anomalies, and execute complex strategies at machine speed. These examples represent only the early stages of a much broader shift toward intelligent automation. As AI systems become more capable and more integrated, they reshape not only how work is done, but how organizations are designed and how value flows through the economy.

The AI economy is poised to surpass the Internet era not only in scale, but in depth of impact. The internet amplified human capability. AI restructures it. The transition from augmenting labor to replacing and redefining it in certain roles carries profound implications for productivity, employment, and competitive advantage.

For organizations, investors, and policymakers, the challenge is no longer whether AI will transform industries. That transformation is already underway. The real question is how quickly institutions can adapt their strategies, structures, and mindsets to operate in an economy where intelligence itself is no longer scarce.

The AI economy represents a seismic shift in how we think about labor, technology, and value creation. By moving beyond traditional IT budgets and into the core of workforce economics, AI unlocks opportunities that far exceed those of the internet era. This is not merely a technological revolution. It is a redefinition of economic possibility.

For those prepared to embrace it, the future is not constrained by legacy structures or linear growth. It is boundary-free, adaptive, and shaped by how intelligently we choose to collaborate with the systems we create.

Disruptive technological waves have always been a fertile ground for innovation, creating unprecedented opportunities for new startups to emerge and redefine industries. Each wave, from the internet revolution to the rise of mobile computing, has not only transformed markets but also allowed a new class of companies to thrive by building natively on the disruptive technology. Today, AI represents the latest transformative wave, presenting vast opportunities for startups that are designed from the ground up to leverage AI’s unique capabilities. These companies, which I call AI natives, stand apart from AI enabled businesses by fundamentally reimagining how products and services are built and delivered.

Lessons from the Internet Era

The internet revolution offers a clear precedent. Consider Netflix and Blockbuster, two companies delivering similar movie experiences at their inception. Netflix, designed as an internet-native company, harnessed the internet as the foundation of its operational model. On the other hand, Blockbuster, an established giant, tried to become internet-enabled by adding digital services to its existing physical rental operations. Despite Blockbuster’s initial scale and market dominance, its inability to fully integrate internet capabilities allowed Netflix to dominate.

This lesson extends beyond entertainment. Companies that build on a new wave of technology from the ground up tend to design their processes, products, and strategies around the unique advantages of the new technology innovations. Retrofitted solutions often suffer from structural limitations, cultural inertia, or slower adaptability.

Disruption creates Opportunities

Disruptive technology waves create opportunities for startups to challenge even the most dominant incumbents. The rise of mobile computing provides a compelling example. Intel, which had long dominated the PC and server markets with its high-performance CPUs, appeared invincible. However, the mobile revolution prioritized power efficiency over raw processing power, an area where Intel’s technology was not optimized. Startups and new players like Qualcomm and ARM capitalized on this shift by designing solutions specifically tailored for mobile devices. These companies leveraged the opportunity to address the new demands of mobile computing, capturing significant market share and reshaping the industry landscape.

Similarly, AI is ushering in a disruptive wave, creating new opportunities for startups to thrive. Traditional CPU models, reliant on serial optimization, struggle to meet AI’s massive data processing requirements. GPUs, with their parallel processing capabilities, have emerged as a superior alternative, giving rise to companies like NVIDIA, which have revolutionized AI infrastructure. This shift underscores how disruptive technology waves open the door for startups to innovate, redefine market standards, and succeed where incumbents may falter.

Rise of AI Natives

AI native startups are built with AI as their core engine, leveraging data, machine learning, and large language models (LLMs) to create differentiated products and services. These companies design their value proposition around the strengths of AI: automation, predictive insights, personalization, and scalability. AI enabled startups, by contrast, integrate AI to enhance existing processes or products. While both categories can create value, their scalability, resilience, and innovation potential often diverge.

Becoming an AI-native startup involves embedding AI at the core of a company’s operations and value proposition. Here are some examples of how companies can achieve this:

• AI-Centric Value Proposition: Build core offerings on AI technologies such as machine learning, natural language processing, computer vision, or generative AI, and focus on solutions addressing problems or creating opportunities that were previously unattainable without AI.

• Scalable and Proprietary Data: Develop, discover and enhance unique datasets that are critical to AI model training and inference to establish strong competitive advantages.

• Automation and Intelligence: AI-native startups aim to automate processes, generate insights, or create novel experiences that go beyond traditional workflows.

• AI Infrastructure: Creating tools, platforms, or frameworks that enable other businesses to implement and scale AI effectively across their operations.

• Market Creation: AI-native startups often create entirely new markets or transform industries by introducing AI-first solutions.

The difference between AI-native and AI-enabled startups is a game-changer for entrepreneurs seeking to thrive in this new era of technology. Just as internet-native companies like Netflix and mobile-native pioneers like Qualcomm reshaped their industries, AI-native companies are leading the next wave of innovation and disruption. For investors and founders alike, embracing this new wave of innovation through an AI-native lens unlocks opportunities to create entirely new markets. By focusing on solutions that harness AI’s transformative power, ensuring scalability, and establishing clear differentiation, businesses can redefine and thrive in an increasingly dynamic industry landscape.

Artificial intelligence did not only arrive as a new technology. It arrived as a new way of computing.

For most of modern history, computation was built on certainty. Early computers were deterministic machines. Given a specific input, they produced a single, predictable output every time. This model aligned perfectly with engineering, accounting, and science, domains where precision and repeatability were essential. Computing was about rules, logic, and control.

That paradigm shaped how we understood machines for decades.

Deterministic Computing: The Era of Certainty

Think about a simple calculator: when you input 2 + 2, the result is always 4. There is no ambiguity, no interpretation, no context. The system follows a fixed set of instructions, and the result is guaranteed.

This is deterministic computing. It excels when problems are well defined, environments are stable, and correctness can be mathematically proven. From spreadsheets to flight control systems, deterministic machines became the backbone of the modern world because they were reliable, consistent, and exact.

But this strength also revealed a limitation.

Deterministic systems struggle because the world is not precise.

When Certainty Breaks Down

Human reality is not binary. We make decisions with incomplete information, shifting context, and competing signals. We reason in probabilities. We revise beliefs. We act without guarantees.

As computing moved beyond arithmetic into perception, language, and decision making, the limits of deterministic logic became impossible to ignore. Many of the problems that matter most cannot be reduced to fixed rules or exact answers. Traditional computing could assist humans, but it could not reason alongside them.

This is where artificial intelligence changed the nature of computing itself.

Indeterministic Computing: A New Paradigm

Indeterministic computing does not aim for certainty. It operates on probability.

Instead of producing one guaranteed answer, these systems generate likelihoods. They infer patterns from data and respond based on what is most plausible. The same input may produce different outputs depending on context, training, or randomness.

This is not a flaw. It is a new way of computing.

Modern AI systems assess the world the way humans do, by weighing evidence, navigating ambiguity, and adapting as new information appears. Language models, for example, do not retrieve answers from a rulebook. They predict responses based on learned statistical structure. Vision systems recognize objects not by definition, but by probability.

For the first time, computing began to resemble human reasoning.

Error, Uncertainty, and Human Likeness

This probabilistic nature introduces a new kind of failure. AI can produce confident but incorrect answers, as we normally call it hallucinations. These moments are unsettling because they break the illusion of machine certainty.

Yet this behavior is deeply familiar.

Humans also fill in gaps. We make assumptions. We misremember. We guess when certainty is unavailable. Our intelligence has always been indeterministic.

AI did not abandon precision by accident. It traded certainty for adaptability.

And in doing so, it crossed a threshold. Computing was no longer just executing instructions. It was interpreting the world.

That shift, from certainty to uncertainty, marks one of the most important transitions in the history of computation.

A Future of Possibilities

The shift from deterministic to indeterministic computing represents a profound change in how we approach technology. As our systems become more sophisticated, they are better able to tackle the uncertainties of the real world. Deterministic systems will continue to have their place, especially for tasks where precision and repeatability are critical. But as we move into a future that requires flexibility, creativity, and adaptability, indeterministic computing will play an increasingly important role in shaping the next generation of technology. The possibilities are vast, and we are just beginning to explore them.

As a technologist by training, I’ve had a front-row seat to the extraordinary transformation in how humans communicate with machines. It has been a remarkable journey, from the painstaking days of programming in machine language to the fluid, intuitive interactions we experience today.

After graduating from electrical engineering in the 90s, my first job was an engineer developing and testing C and C++ compilers on Mainframe which was one of the most in-depth and complex fields mapping the complexity of human intentions into rigid machine instructions. During that time, we had to program in low level machine language, sometimes even writing in binary code, just to communicate with computers and get them to perform tasks. I could picture how read operations move through memory and how write operations change it byte by byte. I could even debug complex triple pointers by mentally mapping the binary ones and zeros, like the streams of code in The Matrix.

In many ways, these early years of computing were like learning a second language. Every command required precise syntax, correct spelling, and a clear understanding of how machines read instructions. There was no room for ambiguity. We could only communicate in ways the machine understood, so most of the burden fell on the programmers. This was the earliest form of “prompt engineering”, carefully writing each line of code so the machine would interpret it correctly.

Graphical user interfaces (GUIs) like Windows 3.0 and the Apple Macintosh were a revolutionary step forward, as they allowed users to avoid command-line prompts and interact more intuitively with their computers with a mouse. Suddenly, we didn’t need to type cryptic commands into a terminal just to get things done. GUIs enabled us to interact with machines visually and intuitively.

In 2007, the iPhone ushered in a new era where interaction became even more natural. Suddenly, the need to memorize commands or navigate complex menus diminished. A user no longer needs to have technical expertise to engage with their device. However, there was still a need to understand which apps to use and how to configure settings, as we adapted our personal technology ecosystems to fit our lives.

With the rise of artificial intelligence, the paradigm has shifted yet again. Today, with AI-powered systems like ChatGPT, we can express our desires in plain language, English or any other ,  and the system interprets, learns, and responds in ways that would have been impossible just a decade ago. AI has become the ultimate abstraction layer between humans and machines, enabling us to skip over the technical intricacies of underlying technologies, apps, or operating systems. Now, the human-machine relationship has flipped; for the first time, humans no longer need to adapt to machines. The machine adapts to us. AI understands the nuances of human language, our intentions, and even our occasional failures to express ourselves clearly. The need to translate our thoughts into “machine speak” is rapidly disappearing, and this shift marks a fundamental evolution in how we engage with technology.

The Implications of a New AI-Native User Interface

We are again on the cusp of another fundamental evolution of a new user interface: the advent of a fully AI-native user interface ,  one where human language itself becomes the programming interface. An AI-native interface could eliminate the need for traditional apps, or even operating systems, as we currently understand them. We no longer need to tell the system what app to open or how to execute commands. Instead, the interface could operate through language, with real-time feedback and response shaping our interaction as it goes.

The future is not just about talking to machines. It is about interacting with them in more natural ways. With AI and machine vision, devices can see what we see and read our gestures, facial expressions, and surroundings. You might wave your hand to clear a notification, use your eyes to scroll, or move your fingers to control a virtual object in augmented reality. This is often called spatial computing. It blends the physical and digital worlds so interacting with technology feels more like moving through real space than tapping on a screen. Instead of typing and clicking, you simply look, gesture, and move. From running your smart home to navigating 3D worlds, you may never need to touch a screen again.

This shift creates interesting investment opportunities: a new kind of interface could emerge, one driven less by apps and more by AI, language, and even body movement. As interfaces become AI native, the line between how humans express themselves and how machines execute tasks starts to blur. Our words and movements become the gateway to the digital world. It is a major leap in computing and a powerful opportunity to invest in the next wave of technology.

A Future of AI-Native Interfaces

The future of user interfaces is simple. Technology should adapt to us, not the other way around. As AI gets better at understanding our words, gestures, and intent, the gap between human thinking and machine logic starts to close.

From a founder’s perspective, this changes what we build. From a technologist’s perspective, it changes how systems are designed. And from a VC perspective, it opens a new platform shift. When the interface changes, entire ecosystems are rebuilt.

We are still early. The winners will not just add AI to old products. They will rethink the experience from the ground up. If we get this right, technology will feel less like a tool and more like a natural extension of how we think and create.

The rise of artificial intelligence did not happen overnight. Nor was it triggered by a single breakthrough or sudden insight. Many of the ideas behind modern AI are decades old. Neural networks, the conceptual backbone of today’s deep learning systems, were first proposed in the mid twentieth century and actively researched throughout the 1980s and 1990s.

The idea was right.
The timing was wrong.

For most of its history, AI existed in the gap between theoretical promise and practical limitation. Researchers understood how machines might learn, but the world lacked the conditions required to make those ideas work at scale. Artificial intelligence was not waiting for intelligence. It was waiting for infrastructure.

Phase I: The Idea Without the Means

Early neural networks were inspired by biology. They attempted to replicate learning through layered connections, weighted signals, and iterative adjustment. By the late 1990s, the mathematics of learning were largely understood. Backpropagation worked. Small models could be trained. In controlled environments, limited success was possible.

But progress stalled.

This era produced one of the most famous milestones in AI history: IBM Deep Blue defeating Garry Kasparov, the reigning world chess champion, in 1997. It was a defining moment, not because the machine “thought,” but because it demonstrated how far computation could go when pushed to its limits.

Deep Blue did not learn.
It calculated.

The system relied on brute-force search, evaluating up to 200 million chess positions per second, guided by human-crafted heuristics and domain expertise. I was there in IBM Watson Research as an architect and developer, building a middleware that allowed the system to scale reliably under extreme computational load. The Deep Blue achievement was extraordinary, but it revealed an important truth: this was not intelligence emerging. It was human optimization amplified by hardware.

Deep Blue showed what machines could do when rules were explicit and the problem space was closed. It did not generalize or adapt. It was intelligence by exhaustion, not by understanding.

The broader field of AI faced the same limits.

Three constraints held progress back.

Compute was insufficient. Training neural networks on CPUs was slow, expensive, and impractical beyond small experiments. Scale was impossible.

Data was limited. The world had not yet digitized itself. Language, images, behavior, and interaction were largely offline, leaving researchers with small, artificial datasets disconnected from real human complexity.

Distribution was weak. Even when systems worked in laboratories, there was no scalable way to deploy them, learn from usage, or iterate in real environments.

AI entered multiple “winters” not because the ideas were wrong, but because the world itself was not yet ready.

The vision of learning machines existed.
The infrastructure did not.

What Deep Blue represented was the ceiling of the old paradigm: intelligence achieved through exhaustive calculation and human-designed rules. What came next would require something fundamentally different, not more clever algorithms, but a world large enough, fast enough, and connected enough for machines to learn from reality itself.

Phase II: The World Digitizes

The first major unlock was not an AI breakthrough. It was the internet, followed by its expansion into mobile life and cloud infrastructure.

I watched this transformation firsthand. At Cisco, during the formation years of the modern internet, we were not thinking about artificial intelligence. We were thinking about connectivity, scale, and reliability. Routers, switches, protocols, middleware. The goal was simple: move information faster, farther, and more reliably. What we did not fully realize at the time was that we were laying the foundation for something far larger than communication.

The internet quietly turned human behavior into data.

Emails replaced letters. Search engines indexed human curiosity. Social platforms captured language, emotion, and interaction at global scale. Smartphones extended the network into every pocket, turning daily life into continuous streams of text, images, voice, location, and behavior. Cloud infrastructure centralized storage and computation, making that data accessible, persistent, and programmable.

From inside the system, it felt like infrastructure work. From a distance, it was something else entirely.

By the late 2000s, humanity had unintentionally created what artificial intelligence had always needed: a vast, living record of how humans see, speak, decide, and interact with the world. Not synthetic datasets. Not laboratory examples. Real behavior, at planetary scale.

Still, data alone was not enough.

Phase III: When Scale Proved the Theory

The turning point came when data and compute finally converged.

GPUs, originally designed for rendering images and video games, turned out to be perfectly suited for the parallel mathematical operations required to train neural networks. What had once been slow and impractical suddenly became feasible. Training times collapsed. Model sizes expanded. Experiments that once took weeks could now be repeated in days or even hours.

This new computational power needed a proving ground. ImageNet became that moment.

ImageNet was more than a dataset. It was a test of scale. With millions of labeled images spanning thousands of categories, it created an environment large enough to expose what neural networks were capable of when freed from data and compute constraints. When deep learning models were trained on ImageNet using modern GPUs, performance did not inch forward. It leapt. The results were not incremental improvements but unmistakable breakthroughs that signaled a permanent shift in how intelligence would be built.

Decades of handcrafted features were rendered obsolete almost overnight. Systems that learned directly from data decisively outperformed those built on human assumptions.

The debate ended.

The lesson was unmistakable. Intelligence does not emerge from clever rules written by humans. It emerges from learning at scale.

Neural networks had always been capable of abstraction. They simply needed enough data, enough compute, and a world large enough to learn from.

Phase IV: Teaching Machines to Decide

If ImageNet taught machines how to see, reinforcement learning taught them how to choose.

Reinforcement learning is not about correct answers. It is about experience. An agent acts, observes outcomes, and adjusts behavior through feedback. For decades, this approach remained mostly theoretical. Real environments were too complex. Training was unstable. Exploration was costly.

Scale changed that.

When reinforcement learning merged with deep neural networks and large scale simulation, machines began learning strategies rather than scripts. They planned across time. They adapted to uncertainty. They discovered solutions no human explicitly programmed.

This marked a second leap. AI systems were no longer limited to perception. They could now learn action, consequence, and adaptation.

Machines began to operate in environments rather than simply classify them.

Phase V: Generalization at Last

With data, compute, perception, and decision-making finally aligned, a final shift occurred.

Researchers stopped building systems designed for single tasks and began training models designed to absorb the world itself. These foundation models were not programmed with rules or logic. They were trained as vast probabilistic engines, exposed to enormous swaths of human language, images, code, and behavior, and asked to predict what comes next.

Language models did not memorize grammar. They inferred it. They did not store facts explicitly. They absorbed patterns of meaning, causality, and context distributed across billions of examples. Multimodal models extended this process across text, images, audio, and action, learning how different forms of information align and reinforce one another.

The breakthrough was about generalization of accessing knowledge

A single model could now adapt across tasks without being rebuilt or retrained from scratch. Writing, reasoning, translation, analysis, design, and simulation all flowed from the same underlying system. Intelligence became reusable rather than bespoke. Learning became cumulative rather than isolated.

This was the moment artificial intelligence stopped being fragile.

AI is the apex of technology evolution

Artificial intelligence did not emerge from a single breakthrough. It emerged when long-delayed conditions finally aligned.

• The core ideas had existed for decades.

• The data arrived as the world digitized itself.

• The compute arrived through GPUs and distributed systems.

• Scale was proven through ImageNet.

• Adaptation came with reinforcement learning.

• Generalization followed with foundation models.

Individually, none of these advances was sufficient. Together, they made rapid progress unavoidable.

What appears sudden is the result of accumulation. Years of theory, infrastructure, and experimentation quietly prepared the ground. Once learning could scale across data, compute, and experience, intelligence could scale as well.

Acceleration was not a surprise. It was the natural outcome of maturity.

Have you ever wondered what our world would look like if the value of a person were measured not by what they do, but by what they create and the insights they offer? Throughout most of human history, with rare exceptions like Einstein or Da Vinci, people have been rewarded more for their productive capacity than their insights. Whether in agriculture, manufacturing, or even knowledge work like programming, compensation has been tied to output. Those with bold ideas but without the resources to turn them into finished products have often gone unnoticed or unrewarded.

For much of history, the ability to live as a visionary depended on others handling the mechanics of execution. In the past, the ability to live as a “Visionary” required the exploitation of other humans to manage the mechanics of life, whether through the forced labor of antiquity or the rigid hierarchies of the early industrial age. Today, AI provides that same execution layer without the moral cost. It separates imagination from exploitation. AI democratizes the aristocracy.

This imbalance did not arise by accident. This dynamic emerged because our major technological breakthroughs were designed to enhance human productivity. Tools like the plow increased agricultural output. The steam engine and mechanized factories unlocked mass production. Digital systems, from spreadsheets to search engines, helped people work faster and with greater accuracy. Each wave of innovation strengthened a model in which technology serves humans as primarily implementers who executed tasks, operated tools, and followed structured workflows. This is the essence of the “implementer economy”.

An implementer economy is based on the execution of well-defined functions. Its value comes from efficiency, consistency, and reliable output. Workers are rewarded for performing tasks correctly and repeatedly, whether assembling a product, processing information, or transporting a passenger from point A to point B. The underlying logic is simple: when output scales only through human labor, ability to execute becomes the driver of value.

The Industrial Revolution embodied this model. Mechanized production, standardized methods, and factory systems turned labor into predictable units of output. The goal was scale. The worker’s role was to operate machines and ensure smooth, repeatable processes. Creativity mattered only if it could be converted into a product. Everything else was noise.

This system powered immense progress. But it also entrenched a world where creativity, imagination, and unconventional thinking were undervalued compared to efficiency and execution. It created an economy optimized for implementers.

A visionary economy is an economic model that values creativity, insight, and invention over routine execution. Its origins can be traced to the Information Age, when computers and the internet first began to reshape businesses and enable more complex, knowledge-driven work. As the 21st century unfolded, the rapid rise of AI and robotics accelerated this shift by automating repetitive tasks and freeing humans to focus on strategy, exploration, and innovation. In this new paradigm, the most valuable human contribution is no longer execution itself, but the ability to imagine, design, and reimagine systems in motion.

This transition does not eliminate the implementer economy overnight. It exists alongside it. But it no longer defines the frontier of progress. That frontier is now driven by a profound technological shift in artificial intelligence. For the first time in history, machines are capable of performing many tasks with a level of autonomy that mirrors human execution. AI systems can process information, make decisions, and carry out actions that once required human judgment. Robotics is evolving into a parallel implementation layer that can move through the physical world with increasing precision. Together, they form a new execution infrastructure, one that operates alongside humans rather than beneath them.

Autonomous driving offers a vivid example. The work of transporting a passenger from point A to point B, once entirely dependent on human labor, is being taken over by robotaxis that operate continuously and with far fewer constraints. Similar patterns are emerging across logistics, manufacturing, customer service, and knowledge work. As machines increasingly handle the procedural and the predictable, humans are called to operate in the realms of creativity, synthesis, and long-term vision.

This shift empowers a new class of contributors. Visionaries who once struggled to turn ideas into reality now have tools that can bring those ideas to life. With AI as a collaborator, individuals can prototype products, test assumptions, and build solutions that previously required large teams or significant capital. The constraint is no longer execution capacity, but clarity of intent and imagination. The combination of visionary thinking and advanced technology is redefining what is possible and unlocking new forms of economic value.

This is the essence of the visionary economy. It is an environment where creativity is no longer constrained by the limits of human execution. It is a world where imagination scales, experimentation accelerates, and insight compounds. We are witnessing a fusion of human judgment and machine capability that is driving innovation at a pace never before seen.

The visionary economy does more than transform the nature of work. It reshapes how we live, collaborate, and create value as a society. The emphasis shifts from execution to intention, from routine tasks to meaningful creation, and from rigid roles to fluid collaboration between humans and intelligent systems.

Where the implementer economy rewarded efficiency and adherence to predefined tasks, the visionary economy values the ability to generate ideas, interpret complexity, and design solutions that cannot be scripted in advance. Insight replaces instruction. Creativity replaces compliance. Judgment replaces repetition.

Collaboration itself is transformed. Teams are no longer defined solely by human expertise. They now include AI agents that analyze information, generate alternatives, and co-create alongside us. The boundary between individual capability and collective intelligence becomes increasingly thin, enabling individuals to operate at the scale of organizations and organizations to move with the agility of individuals.

The implications are profound. Compensation shifts from task execution toward idea generation and strategic direction. Education evolves from memorization to cultivating creativity, systems thinking, and adaptability. Communities organize around shared missions rather than shared locations. And individuals gain new ways to express their potential, no longer constrained by traditional implementation limits.

The visionary economy represents a foundational transformation in how we define progress. It elevates creativity from a personal trait to an economic force. It empowers individuals to create impact at unprecedented scale. As AI and robotics continue to advance, they will reshape not only our jobs, but our lives, relationships, and collective potential.

The future will not be built by those who execute instructions most efficiently. It will be built by those who can imagine boldly, collaborate intelligently, and guide powerful systems with clarity and purpose.

I am not naturally good at expressing ideas.

I learned that early. Growing up in China, I skipped grades multiple times and was almost always the youngest person in the room. The gap showed up physically in gym class and verbally in the classroom. My ideas came quickly, but my ability to express them lagged behind.

So I compensated. I read constantly. I memorized sentences and copied expressions, learning how ideas were formed by watching how others wrote and explained them. Books became my first co-pilots, not because they gave me answers, but because they showed me how to express thought. I was drawn to books about the future, technology, and ancient civilizations, less for their stories than for the patterns in how complex ideas were described. Long before I understood markets or strategy, I was learning by absorbing structure, rhythm, and clarity one sentence at a time.

When I moved to North America at sixteen, the gap widened. I could barely speak English. I slept with the radio on at night for a year, letting unfamiliar sounds repeat until they formed patterns. I started in English as a Second Language (ESL) Level 1 and moved through five levels in my junior year. In my senior year, I finished with the highest academic average across all Toronto high schools, earning perfect scores in mathematics and near perfect scores in physics and chemistry. My English grade was an 80, the only score that pulled down my overall average.

Yet that 80 mattered to me more than the rest, because English was the subject I worked hardest to improve.

It was also a reminder that there would always be areas where I would need help. No matter how hard I worked, language would cost me more effort than it did others. I could improve. I could adapt. But expression would never be my greatest strength.

That reality followed me into my career. After building my own consulting practice and later leading Cisco’s Internet Business Solutions consulting practice in Asia, I regularly advised and presented to senior executives and C-suite leaders at Fortune 500 companies. In those rooms, language fluency was not optional. It was the minimum requirement. As a non-native English speaker, I practiced obsessively. I recorded every presentation. I refined my accent. I repeated each presentation until it felt natural. I did everything possible to remove friction between my thinking and my delivery.

Still, the effort was asymmetric. I was confident in my reasoning and judgment, yet too much of my time was spent translating ideas into words rather than advancing them. I began wishing for something I could not find, a co-pilot that could handle delivery and organization, translating my thinking into expression and words, so I could spend more time doing what I was good at instead of compensating for my constraints.

For years, that co-pilot did not exist. Books helped. Systems helped. Discipline helped. But execution always depended on me doing everything myself.

Then AI arrived.

For the first time, I found an implementation partner. AI did not replace thinking. It structured ideas, refined language, and clarified expression. It gave me back time, not by thinking for me, but by reducing the cost of expression. For someone who had spent a lifetime working around that limitation, the impact was profound.

That is why I am creating this blog.

It is about leverage. I dream about a future where intelligent co-pilots work quietly beside us, carrying the weight of mechanics and translation, so humans can spend more of their energy on curiosity, judgment, and creation. A future where my children do not have to spend endless hours compensating for their weaknesses, but can focus earlier on what they do best, and let AI and robotics complement them where they fall short. If we build these systems with care, then ideas will no longer be limited by who can execute perfectly, but amplified by partnerships that help turn thought into impact. That is the future I hope we can create together.

We are living in a moment of extraordinary transformation. For the first time in history, human creativity and machine intelligence are advancing together, reshaping every corner of our economy, our industries, and our daily lives. What once took decades now unfolds in months. What once required thousands of people can now be achieved by one person with the right tools. And what once separated thinkers from builders is dissolving as Artificial Intelligence (AI) becomes a collaborator that amplifies our abilities rather than replacing them.

Yet the most profound shift is not technological. It is mental. It is a change in how we define value, work, and human potential.

This shift marks the transition into what I call the Visionary Economy: an economic model where creativity, insight, and imagination become the primary sources of value, while execution increasingly moves to intelligent systems. In this world, we are no longer defined by how efficiently we implement tasks, but by what we can imagine, design, and guide into existence. Artificial intelligence and robotics do not replace human potential. They amplify it, acting as collaborators that absorb mechanics and execution so humans can focus on judgment, synthesis, and creation.

The Visionary Economy begins with a mindset. It asks us to look beyond what exists today and imagine what could exist tomorrow. It challenges the assumptions of the implementer era, where value came from efficiency, compliance, and perfect execution, and where creativity was optional. In this new world, creativity is essential. Those who adopt the Visionary Mindset do not see AI as competition, but as a partner that extends human potential and frees us to focus on what only humans can do: imagine, synthesize, and create.

As a venture capitalist, I have seen this shift firsthand. The founders who thrive today are not simply efficient operators. They are builders of new systems, new interfaces, and new ways of thinking. They approach AI not as a tool to optimize existing processes, but as a foundation for entirely new possibilities. Their advantage is not just access to compute or data. It is the mindset they bring to the frontier.

This blog is an invitation to adopt that mindset. It is an exploration to understand the forces reshaping our world and the opportunities emerging at their intersection. Whether you are a leader, a builder, or simply someone curious about the future, the Visionary Mindset will help you see AI not as disruption, but as the beginning of a new creative renaissance.

Because the future does not belong to those who execute the plan.

It belongs to those who imagine what comes next.