Strategy in the Face of Chaos
Audio companions to my writing on strategy, technology, AI, cybersecurity and building technology businesses.
Each edition explores one of my published articles through an AI-generated discussion or debate, offering another way to engage with its central ideas. These are not interviews or original podcast episodes, and the voices are not mine. The written article remains the definitive version.
This channel is currently a pilot, and the format will evolve as I learn what works.
Strategy in the Face of Chaos
Hype as a Coordination Tool and the Trap of “Irreversible Standards” in AI
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
This pilot audio edition explores the argument behind Hype as a Coordination Tool and the Trap of “Irreversible Standards” in AI.
The discussion examines how hype can operate as more than marketing. In fragmented AI markets, a compelling narrative can coordinate investment, talent, partnerships and complementary innovation around a common direction. Using Renaissance Florence as a point of comparison, it considers how narrative itself can become a source of strategic influence.
It also explores the danger that rapid coordination can turn early frameworks, platforms and business practices into de facto standards before their technical or economic merits are fully tested. Network effects, integrations, specialist skills and sunk costs can then make those choices difficult to reverse, reducing competition and locking the market into inefficient paths.
The discussion asks how organisations and policymakers can retain the coordinating benefits of hype while avoiding premature lock-in through open and modular systems, adaptive regulation, careful experimentation and evidence-based adoption.
This is an AI-generated discussion based on my published article. It offers another way to engage with the ideas, but it is not an interview or a recording of me. The written article remains the definitive version.
Imagine you were standing on a massive construction site.
SPEAKER_00Okay.
SPEAKER_01You're the foreman, right. But instead of an architect handing you a precise blueprint for the foundation, your entire crew is just reacting to the loudest person with a megaphone.
SPEAKER_00Oh wow. Sounds chaotic.
SPEAKER_01Right. It's just someone shouting across the site about how incredible tenthouse is going to look. So everyone starts frantically pouring concrete, they're laying steel, not based on a schematic, but based entirely on the enthusiasm of the crowd.
SPEAKER_00Just going off vibes, basically.
SPEAKER_01Exactly. And that megaphone, that is hype. And today our mission for this deep dive is to deliver a really rigorous executive briefing based on an insightful article by Victor Holman.
SPEAKER_00Yeah, it's a fascinating piece.
SPEAKER_01It really is. Because we are going to explore why AI hype isn't just, you know, marketing fluff. It is actually a powerful strategic coordination tool. But simultaneously, it's a potential trap of what Holman calls irreversible standards.
SPEAKER_00Right, which is the scary part.
SPEAKER_01Yeah. So for you, the technology and business leaders listening, this is critical. You need to understand this underlying dynamic before you sign off on your next major AI infrastructure investment.
SPEAKER_00You really do.
SPEAKER_01Okay, let's unpack this. Because we usually treat hype as a dirty word in tech.
SPEAKER_00Yeah, completely.
SPEAKER_01It's like something we are told to look past, something to ignore so we can see the real substance. But Hallman is arguing it's actually the gravity holding the whole AI universe together right now. How does that work?
SPEAKER_00Aaron Powell Well, uh we have to look at the mechanics of an emerging market because in the earliest stages of a technological revolution, there is no gravity. Right. What you have is just chaos. I mean, if you look at artificial intelligence over the last few years, it has been highly, highly fragmented.
SPEAKER_01Troomy different players.
SPEAKER_00Exactly. You have hardware vendors building specialized chips, you've got model developers experimenting with completely different neural network architectures. Trevor Burrus, Jr.
SPEAKER_01Academic researchers publishing papers every day.
SPEAKER_00Right. And regulatory bodies are just trying to understand the implications. Meanwhile, enterprise end users are just trying to figure out how to, you know, automate a spreadsheet.
SPEAKER_01Yeah, they just want a tool that works.
SPEAKER_00And they all operate with completely divergent agendas. So in a market that splintered, nobody can agree on the rules, the technical protocols, or even the business practices. There is no single focal point.
SPEAKER_01So you basically just end up with hundreds of different proof of concepts that never scale because they just can't get the broader ecosystem to align behind them.
SPEAKER_00Yeah, that's the bottleneck. When you lack physical or established economic infrastructure, a powerful narrative has to take its place.
SPEAKER_01The narrative, okay.
SPEAKER_00Yeah. And Victor Holman's article points to this really excellent analogy by Sangjeet Paul Shoudery. He compares this whole dynamic to Renaissance era Florence.
SPEAKER_01Florence. Like in Italy.
SPEAKER_00Exactly. Think about Florence during that time. Unlike its neighboring city-states, Florence lacked typical medieval physical moats or massive geographic defenses.
SPEAKER_01Right. They weren't exactly Sparta.
SPEAKER_00No, they didn't have the hard infrastructure, the massive standing armies or impregnable castles to protect themselves.
SPEAKER_01I want to stop you there, though, because I'm struggling to see how this doesn't just look like standard venture capital investment. Like how is a cultural moat fundamentally different from just having a lot of cash to buy influence?
SPEAKER_00Aaron Powell Because well, cash alone can't force alignment if the ecosystem fundamentally disagrees on the direction.
SPEAKER_01Oh, I see.
SPEAKER_00Florence built a powerful magnetic narrative. They became culturally and symbolically so significant, like the absolute epicenter of the Renaissance ideal, that attacking Florence would impose enormous social and reputational costs on the aggressor.
SPEAKER_01Oh, because you wouldn't just be invading a city.
SPEAKER_00Exactly. You'd be uh destroying the future of art and science.
SPEAKER_01Wow. Okay.
SPEAKER_00They attracted the greatest artists, the brightest philosophers, and yes, the deepest pockets simply by being the place everyone believed was the center of the universe.
SPEAKER_01Aaron Powell That's a massive advantage.
SPEAKER_00It really is. And Holman translates this directly to the AI landscape. A strategic narrative replaces the traditional economic mode.
SPEAKER_01Aaron Powell So if we map that to what we've seen recently with large language models, right? They were bubbling under the surface for years in research labs, but the industry was completely scattered. Completely scattered. Then ChatGPT gains global prominence, and suddenly it's Florence.
SPEAKER_00Yes. It wasn't just a product release, it was a rallying point. It catalyzed immense FOMO fear of missing out.
SPEAKER_01For sure.
SPEAKER_00Capital poured in, but more importantly, highly skilled practitioners immediately gravitated toward the specific models and frameworks behind that tool.
SPEAKER_01Because that's where the momentum was.
SPEAKER_00Right. A whole ecosystem of complimentary services just flourished overnight. It framed this compelling, forward-looking vision that pulled everyone into the same orbit. Hype actually became the coordination mechanism.
SPEAKER_01I still have to push back on this though.
SPEAKER_00Okay, go for it.
SPEAKER_01Because if I go to my board of directors tomorrow and say, you know, our strategy is to follow the loudest megaphone, I am going to be fired.
SPEAKER_00Yeah, they wouldn't love that.
SPEAKER_01Right. So how is relying on hype a defensible business position? Is hype basically just a substitute for actual strategy when an ecosystem is too messy and splintered to regulate itself?
SPEAKER_00Well, it's important to separate following hype blindly from understanding hype as a structural market force.
SPEAKER_01Aaron Powell Okay, make that distinction for me.
SPEAKER_00Holman argues it is the strategy for early stage market coordination at the macro level. Without that hype narrative, major players just cannot effectively pool resources.
SPEAKER_01Because they're all pulling in different directions.
SPEAKER_00Right. Think of it as overcoming inertia. The hype establishes early mover advantages by cementing credibility and urgency. That makes sense. But and this is the crux of Holman's warning for leaders like you listening, while it is a necessary force to rally a fragmented market, it introduces a massive risk.
SPEAKER_01Which is.
SPEAKER_00It risks entrenching early, sometimes heavily flogged solutions as permanent industry standards.
SPEAKER_01Okay, that makes sense. But it ignores the cost factor, doesn't it? We move from the hype to the permanent infrastructure. Like Senghiapal Chadry noted that hype is a trap for late movers, but Holman actually takes it further.
SPEAKER_00Yeah, he looks at how the entire market can get trapped.
SPEAKER_01Right. Because once hype successfully coordinates a fragmented market, that temporary alignment can rapidly freeze into permanent, hard-to-change economic infrastructure.
SPEAKER_00Exactly.
SPEAKER_01We see widely used frameworks become the default standard based entirely on market momentum, not on technical superiority.
SPEAKER_00Aaron Powell If we connect this to the bigger picture, we have to look at the mechanics of why that happens. It's a phenomenon called path dependence.
SPEAKER_01Path dependence.
SPEAKER_00Yeah. The core idea isn't just that past decisions constrain future choices. It's that the cost of switching directions becomes exponentially higher than the benefit of adopting a superior new technology.
SPEAKER_01Aaron Powell Oh, so it just becomes too expensive to change?
SPEAKER_00Right. When an AI narrative inflates, it creates an immediate influx of capital. I mean, look at the valuation benchmarks setting the future baselines right now.
SPEAKER_01Yeah, they are wild.
SPEAKER_00OpenAI is sitting at a $300 billion valuation. Anthropic is at $61.5 billion.
SPEAKER_01Unbelievable.
SPEAKER_00Because they captured the narrative, they captured the resources, and once they capture the resources, the broader ecosystem begins to coalesce around them. Trevor Burrus, Jr.
SPEAKER_01Right. Developer tooling, enterprise workflows.
SPEAKER_00Even third-party software. All of it is built to interface specifically with those market leaders. Trevor Burrus, Jr.
SPEAKER_01So it's literally like pouring concrete.
SPEAKER_00Yes, exactly.
SPEAKER_01Aaron Powell The hype is the liquid state where you can still shape the industry. You can build the forms, decide where the paths go. But once the ecosystem builds tools on top of it, and your workforce learns how to use those specific tools, the concrete sets.
SPEAKER_00Yep.
SPEAKER_01And you're stuck with the driveway you built, even if it's pointing the wrong way.
SPEAKER_00Aaron Powell That analogy holds up perfectly because the concrete sets through network effects. A product or a standard becomes more valuable simply because more people are using it.
SPEAKER_01And some costs accumulate.
SPEAKER_00Rapidly.
SPEAKER_01Yeah.
SPEAKER_00An enterprise doesn't just buy an AI model. They spend millions of dollars training their workforce on how to prompt it. They rebuild their data pipelines to feed it, and they restructure their security protocols to protect it.
SPEAKER_01Well, Victor Holman gives some tangible examples of this outside of AI, just to prove how stubborn these standards can be once the concrete dries.
SPEAKER_00He does, yeah.
SPEAKER_01Tesla's early supercharger network is a great one. They moved fast, they built the infrastructure, and regardless of what other all makers thought was the best engineering solution, Tesla's plug became the de facto standard for EV charging in North America.
SPEAKER_00Right. The switching costs for Ford or GM to build a rival network were just too high.
SPEAKER_01Exactly. But the one that really underscores the absurdity of this, because we literally all touch it every single day, is the QWERTY keyboard.
SPEAKER_00Oh, the QWERTY layout is the textbook definition of path dependence.
SPEAKER_01It really is.
SPEAKER_00We have to remember why it was designed in the era of mechanical typewriters. It wasn't designed for speed or ergonomic efficiency at all.
SPEAKER_01Right.
SPEAKER_00It was designed to slow typists down.
SPEAKER_01Aaron Powell Wait, really? Give me the breakdown on that. Why would anyone design a tool to make the user worse at their job?
SPEAKER_00Aaron Powell Because of the physical limitations of the early hardware. If you typed too fast on a 19th-century typewriter, the mechanical arms carrying the letters would swing up and just jam together.
SPEAKER_01Oh, I see.
SPEAKER_00So the engineers intentionally separated the most commonly used letters, like they placed A under your weaker pinky finger just to pace the typist and prevent mechanical jams.
SPEAKER_01And today, I mean, we are typing on digital glass screens or advanced mechanical switches where a physical jam is completely impossible.
SPEAKER_00Literally impossible.
SPEAKER_01And we have demonstrably more efficient keyboard layouts available. Oh, like the Dvorak layout, which puts all the vowels in the home row.
SPEAKER_00Much faster. Trevor Burrus, Jr.
SPEAKER_01But QWERTY remains the unquestioned global standard because the workforce was trained on it, schools taught it, the manufacturing lines were built for it.
SPEAKER_00Yeah. The cost of retraining the entire world to type slightly faster is just economically prohibitive. The concrete set over a century ago. Which brings us to the critical question for artificial intelligence. What is the QWERTY keyboard of AI that organizations are locking into right now?
SPEAKER_01That is the big question.
SPEAKER_00Holman points out that many organizations are currently treating OpenAI's GPT-based models as the absolute de facto reference point.
SPEAKER_01Right. Everybody's using them.
SPEAKER_00And there might be competing approaches, different neural network architectures that are significantly more efficient, cheaper to run, or less resource intensive. Sure. But those competing approaches are struggling to overcome the massive head start OpenAI has in mind share, developer tooling, and partner integrations.
SPEAKER_01So even if a superior alternative emerges tomorrow, the enterprise infrastructure is already deeply embedded.
SPEAKER_00Exactly.
SPEAKER_01Here's where it gets really interesting, though, because I see where this leads, and it's terrifying from a strategic level.
SPEAKER_00Oh, absolutely.
SPEAKER_01Sangjeet noted that hype is a trap for late movers, right? Because they miss the boat. But based on this, isn't getting locked into a deeply flawed standard early on an even bigger trap for leaders? It is. Because now you are building your entire enterprise on a foundation that might be fundamentally inefficient. You're basically building a massive logistics company based on a fleet of trucks that only turn left.
SPEAKER_00Yeah, it is arguably the most dangerous trap an executive can face. Wow. Holman refers to this as the double-edged sword of hype. When an industry locks into suboptimal standards because it's economically or operationally too costly to change direction, you generate massive economic inefficiency.
SPEAKER_01And how does that manifest?
SPEAKER_00Well, it happens in a few very specific destructive ways. The first consequence is a severe reduction in competition.
SPEAKER_01Aaron Powell Right. Because new entrants can't just build a better product. They have to build a product so revolutionary that it justifies the millions of dollars it will cost an enterprise to rip out all that old concrete.
SPEAKER_00Exactly. They must either conform to the established standard, which you know stifles their unique innovation, or they have to bleed cash to develop an alternative ecosystem from scratch.
SPEAKER_01Which is almost impossible.
SPEAKER_00Right. And Holman highlights a fascinating hardware example in the text. Bolt Graphics and their Zeus GPU.
SPEAKER_01Oh, yeah, I read about this.
SPEAKER_00They are attempting to achieve massive breakthroughs in gaming performance by skipping the traditional process of rasterization entirely and moving straight to full scene path tracing.
SPEAKER_01Aaron Powell Okay, I'm gonna stop you there because I need a 10-second breakdown on what rasterization actually is. Otherwise, I don't really understand what Bolt Graphics is fighting against here.
SPEAKER_00Sure. Think of rasterization as a very clever but very old hack.
SPEAKER_01A hack.
SPEAKER_00Yeah. Decades ago, computers just weren't powerful enough to simulate real physics. So to generate a 3D video game, rasterization essentially takes a 3D scene and flattens it into a 2D image, just guessing where the shadows and light should go.
SPEAKER_01Got it. So it's an illusion.
SPEAKER_00Exactly. Path tracing, on the other hand, simulates actual reality. It mathematically calculates real light rays bouncing around a room. It is vastly superior visually and functionally.
SPEAKER_01So path tracing is just how reality actually works.
SPEAKER_00Yes.
SPEAKER_01But let me guess, the entire $200 billion gaming industry is coded for the hack.
SPEAKER_00The entire graphics industry. All the software, the developer pipelines, the existing game engines, they're all built fundamentally around rasterization. Wow. And Bolt Graphics is pivoting against that standard. Doing so requires an extraordinary amount of effort and resources just to prove their superior method can integrate with a world built for the inferior method.
SPEAKER_01It's a huge uphill battle.
SPEAKER_00They aren't just fighting rival hardware, they're fighting the entire path-dependent history of the industry.
SPEAKER_01And beyond the hardware level, Holman talks about the coordination trap inside everyday business workflows.
SPEAKER_00Yes.
SPEAKER_01This is where it gets really visceral for a business leader. When a company feels that hype-induced urgency, you know, when the CEO or the board demands AI integration by next quarter, stakeholders tend to prioritize rapid adoption over long-term sustainability. They just want the box checked.
SPEAKER_00Which temporarily fixes the short-term fragmentation, sure. But it embeds incredibly flawed solutions into the company's DNA. Let's look at the real-world cost of that. You end up with proprietary data pipelines that don't talk to each other, or half-baked model architectures that are fundamentally insecure or just unscalable.
SPEAKER_01Aaron Powell Yeah. And a year goes by, and the organization has built its daily operations, its customer service routing, its financial forecasting all around these half-baked tools.
SPEAKER_00And extracting them later becomes an absolute nightmare.
SPEAKER_01I can imagine.
SPEAKER_00Imagine having to tell your chief information officer that they need to migrate a live, mission-critical database because the quote unquote hyped proprietary tool you bought last year is now an abandoned ecosystem.
SPEAKER_01Aaron Powell The business disruption is just staggering.
SPEAKER_00It leads straight to the third consequence Holman outlines: economic waste and innovation bottlenecks. Trevor Burrus, yeah.
SPEAKER_01This is a big one.
SPEAKER_00Aaron Powell Markets become overinvested in the narrative rather than the economic fundamentals. All the money flows to the handful of players commanding the hype rather than fostering a diverse ecosystem of potentially groundbreaking projects.
SPEAKER_01Aaron Powell Right. Smaller projects just get overshadowed simply because they aren't part of the dominant hype cycle.
SPEAKER_00Exactly.
SPEAKER_01Have we seen this specific kind of waste before at this scale?
SPEAKER_00Oh, very recently. Holman points to the cryptocurrency and blockchain hype of the mid-2010s as a stark cautionary tale.
SPEAKER_01Oh man, the sheer amount of capital that flooded into blockchain startups just because a white paper mentioned the phrase decentralized ledger was staggering.
SPEAKER_00It really was. To understand the waste, we have to look at what a decentralized ledger actually is. Okay. In simple terms, it's a database that isn't stored in one central location, but is shared and verified across a network of computers. Right. It has very specific niche utilities. But during the hype cycle, it was treated as the solution to everything from supply chain logistics to healthcare records.
SPEAKER_01Right. People were putting everything on the blockchain.
SPEAKER_00It was a misallocation of resources on a global scale. And what's crucial to understand is that despite the considerable inefficiencies, the lack of scalable use cases, and the rampant speculative volatility, the ecosystem's collective momentum persisted long after the initial hype cooled.
SPEAKER_01Because the infrastructure had been funded, the companies existed, and the standard was accepted. Which really highlights just how difficult it can be to dismantle a standard once it has achieved that critical mass of investment. You have massive companies that had to pivot or die because their whole tech stack was built on a hype-driven architecture that didn't actually solve their customers' problem.
SPEAKER_00Yeah, and industries stuck in these traps face a brutal choice. They either have to continually patch their deep-rooted, flawed standards, which creates endless long-term complexity, or they have to undertake a major disruptive pivot.
SPEAKER_01Aaron Powell And doing a massive pivot later, ripping up the foundation of your company's tech stack requires staggering resources and collective buy-in that most leaders just can't muster.
SPEAKER_00No, they usually can't.
SPEAKER_01So for you, the listener, the person who actually has to make these decisions, how do you safeguard your enterprise? Like if getting locked into this proprietary concrete is the ultimate trap, how do we build systems that can actually adapt when the concrete dries? Is it just about picking better software?
SPEAKER_00It goes deeper than that. It requires architectural discipline. Hulman lays out three strategic measures, starting with your primary defense mechanism against lock-in. And that is adopt open modular systems.
SPEAKER_01Modular systems, okay.
SPEAKER_00If you build your internal systems to be modular, you aren't married to a single vendor's proprietary ecosystem. Right. Think of it like using Lego bricks instead of a solid mold. If a better model comes along, a modular architecture allows you to swap it out without having to rebuild your entire data pipeline.
SPEAKER_01That makes total sense.
SPEAKER_00Holman specifically highlights the work IBM has been doing in this space. They're championing open source frameworks and open standards.
SPEAKER_01Which is smart.
SPEAKER_00Right. These inherently encourage healthy competition and interoperability, ensuring resilience and reducing the risk of being held hostage by one platform's dominance.
SPEAKER_01But what if the vendors actively try to lock us in? Because if I'm a massive AI vendor, my goal is to make it as painful as possible for you to leave my ecosystem.
SPEAKER_00Yes, which requires a solution at the macro level. Holman identifies this as risk-based adaptive regulation.
SPEAKER_01Adaptive regulation.
SPEAKER_00Yeah. If one company's hype turns into an inescapable monopoly, it's a disaster for innovation. So regulators have a crucial role to play here, but they have to evolve alongside the technology. Because if they are too rigid, rigid regulation will just freeze the market in its current potentially flawed state. If you try to regulate the algorithm itself right now, you are regulating a moving target.
SPEAKER_01So what do they focus on?
SPEAKER_00Adaptive regulation focuses on the boundaries, specifically incentivizing open interfaces and mandating data portability.
SPEAKER_01Ah, data portability.
SPEAKER_00Yeah. If data is portable, the switching costs for an enterprise drop dramatically. You aren't held hostage. That keeps a playing field dynamic and prevents today's hype winner from becoming tomorrow's mandatory default.
SPEAKER_01Okay, so that covers the architecture and the regulatory environment, but it kind of leaves out the human element inside the boardroom.
SPEAKER_00True.
SPEAKER_01Holman stresses the need to encourage critical assessment, running pragmatic experiments, piloting new tools carefully, and relying on real-world data rather than letting the hype dictate your roadmap. But here is my highly practical question for you, aimed right at our listeners' reality. How do you actually balance pragmatic, evidence-based experiments with the sheer FOMO coming from a board of directors who are demanding an AI strategy by tomorrow morning?
SPEAKER_00Well, it requires a very sophisticated kind of leadership. It's what Holman describes as a balanced way forward.
SPEAKER_01Okay, how does that look?
SPEAKER_00You don't ignore the hype, and you certainly don't try to fight the board's enthusiasm because you will lose.
SPEAKER_01Yeah, absolutely.
SPEAKER_00Instead, you weaponize it. You use the hype internally to overcome organizational inertia. When the board says, we need AI, you use that mandate and that budget to get your teams moving, to break down data silos, and to modernize your infrastructure.
SPEAKER_01I love that.
SPEAKER_00But and this is the critical part, you channel that momentum specifically into open standards and modular systems.
SPEAKER_01But you're basically taking the energy of the megaphone, satisfying the demand for rapid adoption, but you are quietly forcing the crew to follow a modular blueprint behind the scenes.
SPEAKER_00Exactly. You protect your long-term economic viability by ensuring that whatever you build today can be easily dismantled and upgraded tomorrow.
SPEAKER_01That is brilliant.
SPEAKER_00You leverage data-driven validation, running those pragmatic pilots while managing the narrative at the executive level. Hype is a necessary force for moving forward in a fragmented world, but acknowledging its pitfalls is the only way to ensure that your technological progress doesn't turn into a strategic dead end.
SPEAKER_01Which brings us to the core takeaway for you today. Hype is a structural force. It is the gravity that coordinates a fragmented AI market, much like the cultural narrative that protected Renaissance Florence. But your job as a technology or business leader is to harness that gravity without getting pulled into a black hole of irreversible standards. Right. You have to pour your concrete very, very carefully, making sure you are building on open modular frameworks that lets you adapt when the landscape inevitably shifts.
SPEAKER_00And as we consider the sheer speed at which this concrete is currently setting across the global industry, it leaves us with a critical implication to ponder.
SPEAKER_01Let's hear it.
SPEAKER_00If the Corey keyboards of AI are being permanently locked in right now, based purely on today's market momentum and current vendor dominance, we have to ask what future, fundamentally superior AI breakthroughs are we inadvertently suffocating? Wow. Could the ultimate AI winner of the next decade be a company that is actively choosing to ignore today's de facto standards entirely?
SPEAKER_01That is a fascinating question to leave on. Keep your eyes on the blueprint, not just the megaphone. Thank you for joining us on this deep dive. We'll catch you next time.