Table of Contents
Contents
Since the dawn of AI’s industrial revolution, the hands of history seem to have suddenly accelerated. Every few days brings another piece of technology news that is both awe-inspiring and unsettling. Humanity appears to have boarded a high-speed train headed toward an unknown world.
Two recent stories that have sparked intense discussion — one about a digital fruit fly, the other about human cortical neurons connected to electronic systems — along with two podcasts, have given me a great deal to think about. The podcasts feature Tony Kim, Head of BlackRock Fundamental Equities Global Technology, and Eran Zinman, CEO of Monday.com.
The digital fruit fly story concerns a major advance in neuroscience: researchers have successfully converted the complete brain connectivity map of a fruit fly into a functional neural network model, driving the behavior of a “digital fruit fly” in a virtual environment.
In other words, scientists have not only mapped the neural wiring of a fruit fly’s brain — they have begun running that neural network inside a computer, allowing a virtual body to interact with its environment.
The second story comes from a company called Cortical Labs. They have cultivated hundreds of thousands of human cortical neurons and connected them to electronic systems, enabling this biological neural network to interact with a computer environment through feedback loops. The media has sometimes described such systems as a form of “living neural computer.”
On the surface, these two stories come from entirely different directions: one involves cultivating a biological neural network, the other simulating one. But viewed together, they point toward the same technological trajectory:
Intelligence is gradually transforming from a capacity that exists only within biological brains into a system that can be constructed, operated, and replicated.
This reminds me of the sci-fi animated series Pantheon (2022–2023), which deals precisely with Uploaded Intelligence. And, of course, there is the sci-fi classic that has never left my mind: The Matrix.
The possibility that we live in a simulated world seems to be incrementally raised by each new experimental breakthrough, and it is being discussed seriously by an ever-growing number of people.
The two podcasts I have rewatched multiple times both deal with concrete industry developments.
Tony discusses the structure of the technology industry in the AI era (a deeply insightful and brilliant conversation — highly recommended. Watch here: https://www.youtube.com/watch?v=uiYqMl-nwQg).
Eran Zinman shares his vision for the future of enterprise collaboration platforms (watch here: https://www.youtube.com/watch?v=zjcYlEiwnKI).
The digital fruit fly and the connection of human cortical neurons to electronic systems seem to have entered the realm of cognitive science, even philosophy.
At first glance, these two news stories and two podcasts appear completely unrelated. But the more I consider them together, the more I believe they all point to the same question:
Once intelligence begins to be engineered, where does value migrate? And what remains of human uniqueness?
1. When Software Is No Longer a Moat, Where Does Value Go?
For the past two decades, the technology industry has operated under an almost unquestioned assumption: software itself was the moat.
Software was valuable because it was easy to replicate and could scale infinitely. The internet, mobile apps, SaaS — all fundamentally follow this logic.
In the interview I recommend, Tony presents a highly insightful framework. He understands the entire technology industry as a three-layer structure:
- Infrastructure layer: Power, computing capacity, chips, data centers
- Intelligence layer: Models, algorithms, data systems
- Application layer: Software and labor services
Tony uses a vivid analogy to explain these three layers: atoms versus electrons.
Atoms belong to the physical world; electrons belong to the digital world.
Over the past two decades, most of the value in the technology industry has been concentrated in the software and application layer — the top of what Tony calls the tech stack. The enormous market capitalizations of internet software and SaaS companies have essentially been driven by the scale effects of the digital world.
But the arrival of AI is reshaping this structure. As models grow more powerful, code generation becomes cheaper, and the speed of replicating applications accelerates, the scarcity of software itself begins to decline.
Tony offers a critical insight: value does not disappear — it migrates between different layers of the tech stack.
In practice, much of the value once concentrated in the application and services layers is now being pushed downward to the intelligence layer. And the intelligence layer, in turn, depends on the infrastructure layer beneath it.
Some value is shifting toward deeper infrastructure — GPUs, data centers, power systems — the most essential building blocks of the AI era. Other value is migrating toward deeper system architectures.
This naturally raises another question: if software functionality becomes increasingly easy to replicate, where does a software company’s true moat lie?
2. The Real Moat in Software Is Not the Code
In the Monday.com interview, Eran Zinman makes a point I strongly agree with. He argues that the most important source of value for enterprise software in the future will not be any particular feature, but rather the organizational collaboration network.
When people think about moats in SaaS, they often focus on questions like: Are the features more powerful? Is the UI better? Is the automation more advanced?
In the AI era, all of these things are becoming increasingly easy to replicate. What is genuinely difficult to replicate is the way an organization actually operates.
Within a company, every person, every team, every project, every approval chain — and in the future, every AI agent — collaborates within the same system. When a system has gradually accumulated permission structures, workflows, organizational hierarchies, and a history of collaboration, it is no longer merely a tool. It becomes part of how the organization runs. From this perspective, the shift in enterprise software during the AI era is not that “software no longer has a moat.” Rather, feature-based moats are weakening, while system-level moats and network-based moats are becoming more important than ever.
A week ago, I wrote on my social media:
After reading the article “After Losing 80% of Its Market Cap, Monday.com’s CEO Pushes Back on the SaaS Doomsday Narrative: Our Key Moat Is Still Intact,” I went on to watch the full video interview with Monday.com CEO Eran Zinman. The conversation is remarkably candid. I especially agree with the article’s analysis of network effects:
“4. The CEO’s core defensive argument: network effects are still present, and more important than ever.
Among Bustamante’s ten moats, ‘network effects’ is one of the few labeled as ‘still sticky.’ Network effects mean that the more people use the same software, the more valuable it becomes for each user. Bloomberg’s instant messaging feature is a moat not because the chat technology is exceptional, but because every counterparty on Wall Street is on it — you have no choice but to be there too. Eran’s central defensive argument throughout the entire podcast is built precisely on this point.”
What Eran is really saying is this: the future moat of SaaS lies not in software features, but in organizational networks.
BestSign holds a structural opportunity in this wave of AI. In the network-effect-driven contract signing system we pioneered industry-wide in 2018, we have accumulated: enterprise identity systems, permission structures, approval relationships, contract histories, and organizational architectures. Together, these form a trusted contract network.
Many people misunderstand network effects in SaaS, because enterprise software companies (not limited to SaaS) rarely possess genuine network effects. (I have not yet found a particularly strong example — if you know of one, please share.) For instance, Salesforce and many prominent enterprise software companies do not have strong network effects; their barriers come primarily from high switching costs. But Eran believes Monday’s competitive advantage lies in its organizational collaboration network.
Consider a company with 1,000 employees: if every person, every team, and every AI agent collaborates within the same system, that system’s value increases with the number of participants. This is organizational network effects.
Eran’s core argument is that the true moat of enterprise software lies in organizational networks.
The organizational network that BestSign has built actually goes a step further than what Eran describes Monday.com as trying to build. BestSign’s proprietary enterprise trust network extends beyond internal team collaboration. More importantly, it encompasses the mutual trust network between companies and companies, companies and individuals, and individuals and individuals.
If an electronic signature company is merely a contract database, AI agents can operate it. But if the company has built a contract collaboration network, AI actually becomes dependent on that system. The key insight is this: AI can manipulate data, but AI cannot easily replace organization-level collaboration networks and authoritative systems.
This brings to mind a classic passage from Dr. Wu Jun’s Lang Chao Zhi Dian (On the Crest of Waves):
“Over the past century and more, certain companies have been fortunate enough — whether intentionally or not — to ride the crest of a technological revolution. For a decade or so, they come to represent the wave of technology, until the next wave arrives.
To outsiders, everyone at these companies, from the most senior to the most junior, appears to be blessed by the times. For a company, catching one wave may not guarantee lasting prosperity. But for an individual, catching a single wave like this is more than enough for a lifetime. For a young person riding the surf, there is no greater fortune than to catch a great wave.”
3. Can AI Truly Innovate?
As AI begins to write code, author research papers, and design products, a deeper question surfaces: can AI truly innovate?
Today’s large language models already demonstrate extraordinarily powerful capabilities. They can synthesize knowledge, generate new content, and even propose ideas that resemble research hypotheses. But there is a critical distinction here: high-level generation is not the same as genuine innovation.
Tony raised a now-classic question during the interview:
If it were 1904 and AI had only been trained on data up to 1904, could it have proposed the theory of relativity in 1905? If it were 1976, could AI have written the original screenplay for Star Wars? Or today, could it solve the problem of quantum gravity?
Many scientific breakthroughs did not come from computational power alone — they came from someone redefining the problem itself. Newton proposed universal gravitation, Einstein proposed relativity, Turing proposed the theory of computability. The essence of these breakthroughs was not faster calculation, but the introduction of entirely new conceptual frameworks.
If AI can only recombine elements within existing knowledge systems, it will still become an unprecedentedly powerful tool. But it will not replace humanity.
The term “chemical reaction” is wonderfully apt — I have always liked it. It is not only used to describe the spark between people. Many innovations are themselves a kind of chemical reaction: a sudden connection between disparate pieces of knowledge in the brain. Scattered fragments of knowledge, memory, and experience are rearranged and linked in a single moment, and a flash of insight produces a new idea. This is very much what Steve Jobs meant by “connecting the dots.” And these seemingly magical moments are, in fact, built upon the workings of a biologically driven neural network.
Tony mentioned in the interview that his method for learning a new field is to first establish a framework, then fill it in with new information each day. Gradually, these pieces of information connect like a neural network. One day, you suddenly realize: space is connected to semiconductors, semiconductors are connected to energy — and these things form a web in your mind.
I increasingly believe that much of human innovation stems from new connections within this neural network. Different synaptic links create certain flashes of breakthrough thinking.
My personal guess is that this is the hardest thing for machines to replicate — human imagination and creativity.
And precisely because of this, reading and learning — when pursued not for utilitarian goals but out of genuine curiosity — are more likely to produce new ideas. When knowledge from different fields and disciplines unexpectedly converges one day, blending together and triggering a kind of chemical reaction, the results are often surprisingly original.
If the day comes when AI begins to pose questions that no human has ever asked, that will be the true tipping point. Whether that day will arrive, we can only wait and see.
4. What If Consciousness Can Be Simulated?
The digital fruit fly mentioned at the beginning of this article raises an even deeper question. A fruit fly’s brain contains only about 100,000 neurons, yet it still constitutes a complete nervous system.
If we could one day fully simulate a living creature’s brain, what would that mean? One possible interpretation is that consciousness itself may be a computational process. If consciousness can be simulated by a computer, then in theory it could also run inside one.
This leads to a well-worn question:
Could the world we inhabit itself be a kind of simulation?
If we are indeed living inside some form of simulation, then each of us likely arrived in this world with something like a “script.” This also brings to mind several passages from the Jin Gang Jing (Diamond Sutra): “All phenomena are illusory.” “Let the mind arise without dwelling on anything.” “All conditioned phenomena are like dreams, illusions, bubbles, and shadows.”
Perhaps these were a form of timeless warning from the Buddha, reminding people not to cling to the world before their eyes.
This question may sound philosophical, but with the advancement of AI and neuroscience, it is gradually becoming an increasingly concrete technical problem.
As a child, I used to ask one question over and over: what lies beyond the boundary of the universe? No one could give a definitive answer. Later, I discovered that Elon Musk had posed a similar question: if we live inside a simulation, what lies outside the simulation?
I cannot help but wonder — perhaps it is a higher-dimensional space beyond anything we can imagine from within our world, a level where some kind of creator resides.
But here is the real question: even if we clearly recognize that we are living inside some form of simulation, does humanity truly have the ability to escape it?
Perhaps that is exactly why there is that profoundly meaningful saying: “God laughs when humans think.”
One difference between philosophers and ordinary people may be that philosophers are more accustomed to questioning the reality of the world before their eyes.
As far back as ancient Greece, the philosopher Thales posed the question: what is the fundamental substance of the world? From that point onward, Western thought has continually asked: does the world possess some definitive essence? What is the relationship between appearance and reality?
Plato argued that the world of appearances is merely a shadow of the world of ideal Forms, and that true essence resides in the Forms. Aristotle, by contrast, held that essence resides within things themselves, and that appearance and reality are not entirely separate. This divergence eventually developed into two distinct traditions in Western thought: one emphasizing reason and ideal forms, the other emphasizing experience and observation.
The reason no ultimate answer exists is that each advance in science and technology shatters the boundaries of what humanity previously understood.
5. AI Is Changing More Than Just Software
Once intelligence can be engineered, many structures we once considered stable begin to shift.
This is precisely why Tony, in his discussion of AI’s industrial structure, raises a critical question: as intelligent capabilities become cheaper and more ubiquitous, where will value migrate?
Over the past two decades, most of the technology industry’s value has been concentrated in the software layer — the “electrons” of the digital world. But as intelligence itself begins to be mass-produced, the scarcity of software functionality will gradually decline. Value then begins to redistribute across the system: some flows back toward physical infrastructure — computing power, electricity, and data centers. The rest concentrates in deeper system layers — organizational collaboration networks, data architectures, and workflow systems.
In other words, when intelligence is no longer a scarce capability, what becomes truly scarce is not “whether you have intelligence,” but rather: who can build the systems that carry intelligence. Once intelligence is engineered, real value begins to settle within network structures. And once a network is formed, it is very difficult to migrate. In a sense, the network itself is becoming the new system architecture: it both carries intelligence and gradually becomes the new vessel of value.
From industrial structure, to enterprise software’s organizational networks, to digital brain experiments in neuroscience — these seemingly disparate developments are in fact all pointing toward the same trend: intelligence is transforming from a capability into an infrastructure.
The structure of our world is shifting accordingly. At the industrial level, value will be redistributed. At the organizational level, systems and networks will become more important. At the cognitive level, the boundaries of human creativity will be reexamined. And even at the philosophical level, our understanding of reality itself may be altered.
This, perhaps, is where the age of AI truly begins.
We have already boarded this high-speed train. For all of humanity, does each industrial revolution ultimately liberate us, or will it become a new crisis? Is industrial civilization expanding human freedom, or quietly eroding it — perhaps even accelerating humanity’s own end?
Whatever the answer, the train has already departed.
