AI Investment Frenzy Echoes the 1990s Dot-Com Bubble

Companies everywhere are forcing AI down the throats of consumers and workers. The tech sector has reached peak absurdity. A billionaire who has never worked a 9 to5 went from predicting civilization collapse to claiming Chachi PT is essential for raising children. A chipmaker proclaims that virtual AI agents will replace teachers, nurses, and doctors. And Google's CEO declares AI to be an even more profound discovery than fire or electricity. On the other side of the world, a Japanese mogul who lost billions of dollars on wei work and a dog walking app just 5 years ago is now warning that anyone not using AI will be a goldfish. Having exhausted private sector funds, these profits are

now turning to the public sector, pitching AI as a taxpayer funded necessity for Western military supremacy. The greatest tech innovations of the last century, from the microprocessor to the PC, all flourished through organic market demand without fear-mongering or institutional gaslighting. People didn't adopt Wi-Fi, smartphones, or GPS under the threat of extinction. They adopted them because they worked. Their inventors never needed to pose as doomsday profits. Supply trail demand products evolved through iteration and their mass market impact was earned, not declared. Today, that logic is inverted. Billions are poured into scaling a technology that has stagnated and is still searching for

its form factor. AI is a narrative manufactured by tech billionaires and echoed by the politicians chasing an economic silver bullet. Yesterday's crypto and NFT evangelists are today's AI champions. The AI bubble of 2025 is not a carbon copy of the dot bubble. Yet, the parallels are undeniable. In software, you have an unprofitable first mover who has timing but no moat and a swarm of VC backed startups trying to colonize every market with the same platform. In hardware, there's a shovel seller who's extracting billions, selling a global dream, and pushing infrastructure for demand that hasn't materialized. And behind them is a presidency that has latched onto this technology as a way to project American exceptionalism, blunt Asia's

manufacturing dominance, and mask economic cracks at home. It's different technology, but the same house of cards disguised as a free market boom. Today's large language model wars are a beat-forbeat remake of the 1990s browser wars. Open AI is the new Netscape, a first mover who's burning cash with no lasting means to capture the value that they create. Nvidia is Sun Microsystems and Cisco combined. Today's hyperscalers are Exodus Communications, spending billions to build capacity for a revolution with no viable unit economics. In this modern MBA episode, we deconstruct this house of cards that is the AI bubble, peel back the layers to expose the companies and fragile dependencies within this tech stack, and

reveal why today's $500 billion frenzy is a near identical echo of the dot bubble 30 years ago. The.com bubble is significant not just because of its similarities to today's AI bubble, but also because of how it's shaped the tech sector as it exists today. The fallout in 2000 taught venture capitalists a smarter way to pump and dump. Instead of rushing to IPO, VCs now keep startups private, using series C, D, E, and sometimes F rounds to supercharge growth, control narratives, and inflate valuations through backroom deals within a closed circle of elites, journalists, and fund managers. By the time these tech startups IPO, their growth curves are already spent, future potential is

already priced in, and the burden of profitability is dumped onto the public. While the VCs and founders get to exit quietly with life-changing fortunes, today's AI startups pursue war chests that make the 2010s blitz scaling look like seed rounds. Their media appearances and keynotes are choreographed to sustain hype, and the VCs behind them play the same games, evangelizing their other portfolio companies and coordinating IPOs behind the scenes, so all their investments can exit at peak valuations. As a result, the AI bubble is more bloated than anything else in tech history, surpassing even the.com bubble when adjusted for inflation. The current landscape is as raw as it is saturated.

Thousands of startups have raised millions of dollars for a seat at the table. Even though all the wealth pulls around the first movers from BTOC rappers and enterprise agents to model repositories and middleware, every market is crowded with lookalike AI startups. Because everyone is building the plane while they're flying it, differentiation has devolved into a war of benchmarks where the only moat is a marginally better column graph. Ultimately, the only way to draw parallels between the AI bubble today and the dotcom bubble 25 years ago is to understand the entire tech stack from hardware to software. Like web van and pets.com during the dotcom bubble, the

highest, frothiest, and most visible layer are the consumer AI startups. They boast the most radical value props, generate the greatest hype, and burn most of their money on marketing just to sustain their astronomical valuations. These startups are app layer tenants built on the platforms below them. If these underlying platforms raise their prices, restrict their API access or integrate features natively, then the modes of these apps on top vanish. This fragility is not just for the B TOC, but also the various B2B startups in legal, healthcare, coding, and support. For them, the reality is the same. It's high burn, no margins, no edge beyond timing, and zero control over the large language

model platforms that enable their existence. Like Webban and Pets.com in the '90s, this layer will always be the first domino to fall. These startups burn record cash to outrun not only commoditization but also their own underlying platforms under the hope that their data, UI or brand will somehow be enough to stop customers from defecting to cheaper native bundled alternatives when they inevitably arrive. Below the application layer sits the software gateways. These are the big bang catalysts that drive the innovation as much as economic excess. In the '90s, the web browser was the match that lit the fire. Netscape pioneered the graphics-based web browser, transforming the internet from a textonly academic

ghost town into an interactive digital frontier with encryption, progressive rendering, and JavaScript. Within a year, millions had flocked to Netscape as the front door to the internet. Their success invited a swarm of fast followers, ignited the.com bubble, and culminated in a landmark IPO for Netscape. As the first mass market gateway to the internet, Netscape used their user growth, technical achievements, and first mover advantage to justify their staggering losses and record valuations. Today, it's Open AI who has pioneered the first mass market large language model with chatbt as the gateway to generative AI. They justify their valuation and burn rate with their user count and market timing. Just like

in the 90s with Netscape and web browsers, LLMs have quickly become a crowded field with many fast followers who have replicated OpenAI's technology and products. A tech stack is a hierarchy of dependencies with each layer building on the one beneath it. While software gateways serve millions of users, they remain tenants to the operating, distribution, and device platforms below them. These foundational systems control the hardware, OS, and access points. In the browser wars of the 1990s, this layer was simple. Netscape was tethered to the desktop PC where Microsoft Windows held a monopoly. In today's large language model wars of the 2020s, independent LLMs similarly remain at the mercy of their hosts.

ChatGpt, Claude, and Perplexity are guests that reach users through the browsers, app stores, hardware devices, and search engines owned by Apple, Microsoft, and Google. They're unable to penetrate the walled gardens of big tech where the most valuable user data and critical tasks reside. This is why Meta has been the most aggressive incumbent in going allin on AI. To Zuckerberg, AI represents a generational opportunity for Facebook to finally break its dependency on Apple and Google. He's racing to turn LLMs into some kind of breakthrough device just to secure the platform ownership that he's pursued for over a decade. Beneath these layers lies the proprietary hardware that enables

these software gateways and this innovation. In the9s, Sun Micros systemystems and Cisco dominated, selling high performance servers and routers essential for docom scaling. You could not build an online business without either. Only Sun servers could handle the heavy processing loads, and only Cisco's switches could route the traffic. Both enterprises made fortunes selling docom shovels as companies all around the world rushed to scoop up their hardware. Meanwhile, the US government tightly regulated exports, classifying Sun servers as national security assets and blocking sales to rival countries under the concern that this advanced computing power could fuel military and intelligence advancements.

30 years later, this exact same pattern is now playing out with AI. Today, you can't build a Frontier LLM without Nvidia's GPUs. Because LLMs require constant training for accuracy, every player in the stack is trapped in a neverending arms race. They must have access to the most GPUs possible, both in power and quantity to maintain their competitive edge. Nvidia continues to post record sales quarter after quarter as companies all around the world buy out their latest GPUs before they're even released. Nvidia today is Sun and Cisco combined. They own the hardware, networking, and programming. And just like in the '90s, the US government has placed Nvidia's GPUs under some of the

strictest export restrictions in order to maintain American hijgemony. But hardware requires a home. In the9s, Exodus Communications was the dominant infrastructure provider, hosting the Sun servers and Cisco routers that powered Google, eBay, and Yahoo. As do startups entered the market, they funneled their venture capital funds to Exodus as rent, which Exodus then used to build even more capacity. When the dotcom bubble burst in 2001, these startups collapsed. Their payments disappeared and Exodus went bankrupt with a sprawling empire of unused hardware and empty data centers. Today's infrastructure layer is more accessible, but follows the same dynamic. While the do startups purchased

and own their own Sun servers and Cisco routers, AI startups today virtually rent Nvidia's GPU on demand from Amazon AWS, Microsoft Azure, and Google Cloud. These big public cloud providers own, host, and rent out millions of GPUs, CPUs, and disc space around the world. These tech giants have spent decades competing with one another to be the ultimate landlord of cloud computing. The AI frenzy has not only lifted up these big three public cloud providers, but has also birthed Neoclouds like Coreweave, whose entire business model relies on exclusively hoarding and renting out the latest and highest end

Nvidia GPUs. What makes this layer so fragile is that while startups stockpiled Sun servers in anticipation for future growth and traffic that never materialized, today's AI startups are clinging on to the unproven brute force scaling laws that throwing more GPUs and more data will yield in a more intelligent model. But unlike the startups who purchased hardware to scale, these AI startups are renting hardware to simply survive. It's a dangerous cycle. Startups burn their venture capital to rent GPUs from these cloud providers. And the cloud providers use that money to buy more chips from Nvidia and to build more data centers.

It's a ridiculous loop that's sustained solely by the speculation that infinite scaling will eventually yield a profitable product. At the base of the stack is utility. In the '90s, the bottleneck was bandwidth. While Sun provided the server side power, it couldn't fix what happened on the client side. Users surfing the web face constant congestion, timeouts, and sluggish loading. Legacy telephone copper lines could not handle the millions of users and massive data packets that this new emerging web demanded. This physical constraint sparked fears that the dot industry would collapse, and it triggered an arms race within the private sector to build this new digital railroad for the 21st century. The promise of high margin

reoccurring utility revenue pushed telecom companies to spend trillions and borrow billions to lay millions of miles of fiber. However, when the dotcom bubble burst, these giants like WorldCom and Global Crossing collapsed, having overleveraged themselves to build for a future that had not yet arrived. Today's bottleneck for AI is energy. Modern chips are far more power intensive than the sun servers of the '9s. The public grids can't supply electricity to run and cool Nvidia's GPUs at the pace that the industry is demanding. A single AI data center can consume as much power as 100,000 households. With public grids unable to scale, the big tech companies are aggressively investing in nuclear

energy to hoard the electricity required for their AI workloads. Companies are obsessed with AI for its promise to lower costs, and it's only made it harder to speak to a real human. They'd rather connect you with chatbots whose only purpose is to waste your time and save them money. Everyone is building AI for shareholders and no one is building for the consumer. At Modern MBA, we're the biggest AI critics. When Kudos approached us saying their AI could lower our internet and phone bills without us having to pick up the phone or write a single word. It sounded like Silicon Valley vaporware. Then we tried it ourselves. Kudos successfully negotiated down our Xfinity bill by 22%

and increased our download speeds by 10%. Hello. Thank you for calling in. This is right here and I have your account number. Hi. Thanks for taking my call. My account number is and I'm calling because I'd like to see if I can lower my monthly internet bill. Sure thing, sir. Let me uh pull up your account and uh were you able to find any promotions for me? That could lower it down for $95.

Your speed will be upgraded to 1,200 Mbps and at the same time it says 95. The gun has been applied. Your speed has been upgraded. Everything is all good from my end. Thank you. Have a great day. Kudos is AI that fights for you, the person paying the bill rather than the companies collecting it. No more hours wasted on hold. No more grumpy agents. Just money and time saved on your biggest expenses. Kudos takes the fight to all the service providers and credit card companies who have buried consumers with legal ease, deceptive UI, policy traps, and offshore call centers. For

decades, these corporations have made it intentionally hard to talk to a real human so that customers give up trying to cancel or negotiate their bills. That all changes with Kudos. Kudos's AI voice agent is so natural that not even your Verizon, Comcast, or Direct TV rep will know its AI. The results speak for themselves. The average Kudos user saves over $250 per provider per year. There's no contest. Rocket Money charges you $250 per year, takes 50% of every dollar saved, and drags negotiations out for nearly 2 weeks. In comparison, Kudos saves you the same amount in 10 minutes, and you keep 100% of all those savings in your pocket. Negotiating your bill is just the beginning. Kudos's voice agent

will soon help you dispute charges, secure the best retention offer, and you'll never waste a single second on hold. It's not just AI. Kudos also cancels unwanted subscriptions in seconds, maximizes benefits on every purchase by picking the best card from your wallet, stacks rewards you didn't even know existed, and offers cash back at over 15,000 websites at rates even higher than Rocku 10. Download Kudos today for free and let AI fight for you, not the corporations. Kudos is available on iOS, Android, and in your favorite browser. Modern MBA viewers get a $20 cash bonus after their first confirmed purchase on a Kudos partner website with the referral code Modern MBA. Thank you to Kudos for fighting for consumers and

making this episode possible. The money flows from top down. When venture capital, speculation, and demand dry up at the application and gateway layers, this entire house of cards will collapse. Ultimately, open AI is not a onetoone for Netscape. Just like Nvidia is not a onetoone with Sun. Unlike their do predecessors where there were only one or two players on every layer, today's tech giants are diversified walled gardens who own across the entire stack from the OS and browsers to the hardware itself. The executives, boardrooms, and venture capitalists that lead big tech today are the tech workers from the 1990s who forged their careers in the dotcom fallout. They're all

extremely sensitive to being abstracted away. And that's why these big tech companies maintain armies of engineers so they can institutionalize any emerging technology. They hire these massive account teams to lock in Fortune 500 clients. And they fund these giant M&A divisions that are ready to absorb promising startups before they ever become full threats. And in Silicon Valley, no CEO wants to be remembered as the one who screwed the pooch, like Marissa Mayor at Yahoo, Carly Fiorina at HP, Steve Balmer at Microsoft, or John Scholley at Apple. This is why Meta, Google, Microsoft, and Apple are all forcing AI into every product and service without a clear user need or

endgame. They would rather subsidize adoption and accelerate this uncertain future towards commoditization than give up an inch to a newcomer. With their walled gardens, these tech giants can afford to burn billions of dollars. In the worst case scenario, AI will just be another write-off for big tech, no different than the Apple car, the metaverse, the Windows phone, or Google's various failed products. From this angle, their collective spending spree is an insurance policy against a total platform shift. If LLMs become their primary touch point, then a new AI hardware device could supplant the iPhone. If people stop Googling, then Alphabet loses its search monopoly. If AI agents take over, then the need for

companies to buy Microsoft Office disappears as the underlying Excel, Docs, and PowerPoints just become invisible. The simplest way to understand is that if AI really does become the primary way that humans interact and work, then the billion-dollar interfaces that make big tech big in the first place risk becoming low value invisible components. Just like how web browsers eventually became free integrated offerings, big tech is betting that LLMs will follow the same pattern as commodity add-ons rather than independent standalone proprietary services. Ultimately, it's a matter of when and not if the dominoes fall, and the sequence is clear. Every participant has actively inflated this bubble for their own personal

enrichment, but it's open AI that warrants the most scrutiny. Sam Alman is the architect of the AI bubble, having inflated breakthrough software innovation into a trillion dollar global narrative with capital raises and promises that dwarf any in Silicon Valley history. His rhetoric has emboldened hyperbole, secured sectorwide immunity for IP and copyright theft, and gifted the political ruling class a convenient diversion from domestic issues. This trillion dollar house of cards has been forged through backroom deals, political favoritism, and circular financing, which has now bound every company in the stack to the same shared fate. But beyond Sam Alman's personal motives, Open AI is at a breaking point. Innovation is no

guarantee of economic success, and a massive user base is meaningless without a path to profit. Open AI's position today is both enviable and untenable. It's a mirror of Netscape in the 1990s, a market-making category defining pioneer that's destined to be swallowed by the underlying platforms that it depends upon. Netscape was the chatbt of the dotcom bubble. It revolutionized how people interacted with the internet, ignited the browser wars, and leveraged a premium model to grow to a 90% market share. But while innovation fueled its meteoric rise and historic IPO, it couldn't pay the bills. Netscape eventually died because it failed to

build a viable business before the walls closed in. It was Microsoft who delivered this fatal blow. Bill Gates realized that the web was more engaging than the Windows desktop. And he worried that the more time users spent in a third-party browser, the less they would value the underlying operating system. If developers began to prioritize Netscape over Windows, then his empire was dead. Microsoft retaliated by bundling Internet Explorer as the pre-installed default and forcing OEMs to exclude Netscape. So, while Netscape still retained its user base, it was effectively choked off from growth and unable to compete with the convenience of native distribution without exhausting its own capital. Before its

collapse, Netscape tried for years to transition from a breakthrough product into a sustainable business. Like ChachiBT, it had a premium model, but it rarely enforced its licenses out of fear of slowing user growth. Netscape tried to be an enterprise super browser, a kind of unified B2B interface for email and reports across all operating systems. But Microsoft and IBM quickly boxed Netscape out by bundling the same services for free in their own software suites. By 1998, Netscape was desperate for cash flow, and it pivoted back towards consumer advertising. Yet, it was all too late. Internet Explorer had cannibalized its market share as the built-in alternative, while Yahoo and

AOL had already secured dominance upstream as the primary ad hubs for the internet. Facing bankruptcy, Netscape sold itself to AOL, another independent gateway that was seeking platform ownership. Ironically, the web browser that kickstarted the dot bubble never lived long enough to see its peak, and AOL was eventually crushed by the same forces that swallowed Netscape. AOL's business was built entirely on legacy telephone lines. Once the cable company started building high-speed broadband, AOL's dialup business evaporated, and eventually so did their ad business as users abandoned the AOL portal for the freedom of Google search and the open web. Yet, the parallels between Netscape

and Open AI extend into politics. Just like how the Trump administration has anchored its strategy to AI, the Clinton administration architected the information superighway through the 1991 high performance computing and 1996 telecommunications act. The federal government provided legal protection, waved regulation for dotcom companies and mandated Netscape as the browser of choice for millions of students and government employees. At the same time, the Clinton administration advocated for Netscape browsers, Cisco switches, and Sun servers abroad so that the US could dominate this emerging web. And when Microsoft began bundling Internet Explorer for free into Windows, the DOJ

launched an antitrust lawsuit as a taxpayer funded attempt to protect Netscape. 30 years later, the US government is running the exact same play with Open AI. The AI litigation task force and Genesis mission grant Open AI legal immunity from copyright liability, IP infringement, and responsibility from AI generated output. And this protection in theory extends to all AI startups. At the same time, Nvidia has effectively become a branch of American foreign policy through the Chips Act. Just like how the Clinton administration brokered international deals for Sun and Cisco, the current administration is doing the same and using taxpayer funds to buy GPUs in the name of public interest. All of which

further enrich Nvidia. Ultimately, AOL and Netscape were the first movers that defined the mass market but failed as businesses. They were independent gateways where users merely passed through. The $60 billion that Sam Alman has raised only buys him time to find a moat. The company's latest moves and internal panic over Gemini 3 says everything. Apple, Google, and Microsoft are all using the Internet Explorer playbook to suffocate open AI. For them, it's a war of attrition. When Apple intelligence on an iPhone, Copilot in Excel, or Gemini in Google Docs becomes good enough for the average user, the convenience of this free built-in default will always defeat performance

and first mover advantage. And since these titans can indefinitely absorb these massive losses, Open AI must somehow defend their paying subscriber base without running out of money themselves. As time passes, they have less and less leverage. If Open AI raises prices or falls behind on performance, then chat GPT customers will simply turn towards the nearest alternative. Any delay to GPT6 or GPT7 will allow incumbents to close the gap. And in this self-made bubble, Open AI can't afford to be silent. It has to ship things constantly, even if it's only for optics, to avoid the perception of falling behind. From this angle, AI agents are less of a technical epiphany and more of a calculated attempt to puncture the walled gardens of big tech

without triggering an all-out war. By using agents to remotely book flights, manage calendars, and order food through what simply appears on a user's screen, Open AI and other AI startups can bypass the API restrictions, platform policies, and harvest the user data they desperately need without violating the terms of service of these big tech companies. At the end of the day, all Open AI can do is to throw everything at the wall to find a mo before this window slams shut. They've spent over6 billion dollars to bring on Johnny IV in the hopes of designing a device that can disrupt the smartphone. They've launched a browser, embedded texttovideo generation, created their own app store,

and release search GPT shopping to facilitate online transactions. Yet, their biggest bet is in physical infrastructure. The taxpayer funded $500 billion Stargate would be the world's most powerful supercomput. Their hope is to create a computing monopoly that's so strong that even big tech will be forced to rent their hardware to keep up. Ultimately, it's only a matter of time before Open AI will be forced to monetize their gateway directly where chat GBT responses will be mixed with ads and user data gets auctioned off to the highest bidder. This situation is not unique to open AI. In the com era, Opera was the underdog browser. Its pioneered speed, tabs, session recovery,

and pop-up blocking. But despite leading on innovation, ad revenue was never enough, and Opera was still eventually obliterated by Internet Explorer. In this AI bubble, as is in the9s, convenience and distribution are the only modes that matter. And Anthropic faces the same issues as open AI. Just as Netscape discovered the hard way, Open AI today is realizing that general purpose gateways grow the fastest, but are the first to die out from commoditization. They are the hardest to defend and sustain as independent businesses. But outside of the numbers, Netscape was ultimately not wrong. They viewed the internet as a paradigm shift that would democratize commerce and launch a software revolution. All of which have come true 25 years later. The

only distinction was that Netscape was not the right company, nor did it survive long enough to realize the future it had made possible. Sun Micros Systemystems envisioned a world where the network was the computer where data would reside online and that computing power would be sold like electricity from a grid. Today, we live in that reality. We access our data across multiple devices and we rent compute from the cloud. Cisco saw a world where voice, video, and data would converge together and that workers would be liberated from cubicles and desktops. All of these pioneers were right about what would happen, but they could not have known how. Ultimately, for all of

these predictions, it took Apple to provide the missing piece in the smartphone to turn these projections into reality. From this lens, open AI may be right about LLMs being some pathway to AGI, how they'll transform industries, unlock greater productivity, and redefine human purpose. But like the tech pioneers before them, they're trapped in an economic bubble of their own making. If it really does exist, the AI revolution, very much like the internets, will take decades and require even more breakthroughs that no single startup can achieve. Yet, the hard truth is that this bubble is not necessary. By framing AI as some existential techno theological crisis, Open AI and all the other AI startups riding on this

narrative have effectively socialized the risks while privatizing the speculative economic gains for themselves. When the AI bubble finally bursts, it'll be the public that bears this wreckage while the circle of tech billionaires and political elites continue to profit, free to manufacture the next tech bubble to fuel their fortunes and power. Every business is asking the same question. How do we make AI work for us? The possibilities are endless and guessing is too risky. Here's the problem. While you're trying to figure it out, the competition is already moving. And every day you sit on the sidelines, you're losing ground.

Forget the buzzwords and snake oil pitches. With Netswuite, you get real, actionable AI that actually gets your business. No fluff, big zany words, just results you can use literally tomorrow. Netswuite is the number one AI cloud ERP trusted by over 43,000 businesses. Netswuite is a unified suite that brings your financials, inventory, commerce, HR, and CRM into a single source of truth. Even great instincts only get better with data. By connecting all the data in Netswuite, your AI gets smarter. It doesn't just guess, it knows. Netswuite intelligently automates routine tasks so you can deliver real value, cut costs, and make fast AI powered decisions with confidence and

total flexibility. Now, with Netswuite AI connector, you can use your favorite AI to connect to your actual business data and ask every question you've ever dreamed of, from key customers and cash flow to inventory trends and accounting projections. Plus, it'll automate all your tiresome manual processes. Let's see your competition do that. Netswuite isn't another catchy rapper. It's AI built into the system that runs your business. Our goal at Modern MBA is to grow to seven figures. And once we reach that milestone, we plan to adopt Netswuite. We sit on a mountain of data, but production is so intense that no one has the bandwidth to analyze the business. We're constantly balancing

creativity with economics, but it's hard to derive without pulling ourselves away from production. If your revenues are at least in the seven figures, get the business guide, demystifying AI, free at netsweet.com/modernba. The guide is free to you at netswuite.com/modernba.