In May 2012, I was at IG Group when Facebook went public at $38 per share, valuing the company at $104 billion. We had clients on both sides of the trade — some convinced Facebook was worth every penny, others certain it was the most overvalued IPO in history. The share price dropped 50% within four months. The consensus was swift: Facebook was a bubble.
They were wrong. Not wrong about the initial mispricing — the IPO price was indeed too high for the company's demonstrated earnings at that moment. But profoundly wrong about the underlying value. Facebook's real worth was not in its platform or its revenues. It was in a portfolio of intangible assets the market could not see: a data asset of unprecedented scale, network effects that strengthened with every user, advertiser relationships that would eventually generate hundreds of billions in revenue, and an engineering culture that would produce Instagram's acquisition and WhatsApp's integration — both of which initially looked like overpayments and subsequently proved to be among the most astute acquisitions in technology history.
The pattern is always the same: markets overvalue the technology and undervalue the intangible assets. Today, in the midst of the AI investment surge, the same pattern is playing out — but with a twist that makes the stakes considerably higher.
At IG Group, we had a front-row seat to market sentiment around the Facebook IPO. Our clients were trading on the share price, and the debate was fierce. The bears pointed to declining user growth in developed markets, mobile monetisation uncertainty, and a price-to-earnings ratio that defied conventional metrics. The bulls pointed to user engagement, platform stickiness, and the sheer scale of the audience.
Both sides were asking the wrong question. They were debating the platform when they should have been valuing the assets.
Facebook's intangible asset portfolio in 2012 included at least five categories of extraordinary value:
Data assets. Facebook possessed the largest structured dataset of human social connections, preferences, and behaviours ever assembled. This data was not merely stored — it was continuously enriched by user activity, making it more valuable with every interaction. Under IAS 38, none of this appeared on the balance sheet.
Network effects. Each new user made the platform more valuable for every existing user. This is the textbook definition of a positive network externality, and it created a competitive moat that proved impenetrable for over a decade.
Advertiser relationships. The self-serve advertising platform Facebook had built was a relationship asset — millions of small and medium businesses learning to use the platform, building audiences, developing expertise. The switching costs were not contractual but cognitive and operational.
Engineering culture. Facebook had built a product engineering culture that valued speed, experimentation, and data-driven decision-making. This organisational capital was invisible to investors but proved to be the engine behind every subsequent product success.
Acquisition capability. The combination of cash, talent, and strategic clarity enabled Facebook to identify and acquire complementary intangible assets — Instagram for visual attention, WhatsApp for messaging graph, Oculus for future platform positioning.
The market did not misprice Facebook because analysts were incompetent. It mispriced Facebook because the valuation frameworks available could not capture intangible asset value. The same frameworks are being applied to AI companies today, with the same structural blind spot. Markets remain largely unable to value what they cannot see on a balance sheet.
Facebook was not an isolated case. The same pattern — overvaluing the technology, undervaluing the intangible assets — has repeated across every major technology wave.
| Bubble | Peak Overvaluation | What Was Overvalued | What Was Undervalued | Resolution |
|---|---|---|---|---|
| Dot-com (1999-2000) | Nasdaq 5,048 (March 2000) | Internet traffic, "eyeballs," page views | Customer data, brand trust, operational infrastructure | Survivors (Amazon, Google) won on intangible assets, not traffic metrics |
| Social media (2012-2013) | Facebook $104B IPO | User counts, platform features | Data assets, network effects, advertiser relationships | Market corrected downward, then massively upward as assets were monetised |
| Crypto (2017, 2021) | Bitcoin $69K (Nov 2021) | Token prices, protocol novelty | Smart contract infrastructure, developer ecosystems, institutional tooling | Survivors building durable infrastructure assets retained value |
| AI (2024-2026) | OpenAI $500B, Nvidia $3T+ | Model capability, benchmark scores | Proprietary datasets, fine-tuned models, AI-augmented workflows, organisational AI capital | In progress |
In every case, the market initially valued the visible technology — the website, the app, the token, the model — and missed the invisible assets that determined which companies would create lasting value. The correction, when it came, did not destroy the technology. It repriced the companies based on their actual intangible asset portfolios.
The AI investment landscape in 2026 exhibits every characteristic of a technology bubble: exponential valuation growth, consensus narrative around transformative potential, massive capital inflows, and a pervasive inability to distinguish between companies that are building durable value and those that are riding momentum.
OpenAI tripled in valuation from $157 billion to $500 billion within a year. Nvidia's market capitalisation exceeded $3 trillion on the back of AI chip demand. Hundreds of AI startups have raised at valuations predicated on projected AI revenues that have not yet materialised. The capital flowing into AI is unprecedented — and much of it is being allocated without any rigorous assessment of whether the underlying companies are building intangible assets or merely consuming them.
The AI bubble is not uniform. Some AI companies are genuinely building extraordinary intangible asset portfolios — proprietary training data, fine-tuned models, AI-augmented organisational processes. Others are API wrappers with marketing departments. The correction, when it comes, will devastate the latter while rewarding the former. The ability to distinguish between them is the central challenge for investors in this market.
How do you tell the difference between an AI company that is genuinely valuable and one that is a bubble? The answer is the same in 2026 as it was in 2012: look at the intangible asset portfolio, not the demos.
A genuinely valuable AI company builds assets that compound. The five questions that matter:
Does the company own proprietary training data? Companies that train models on proprietary datasets — data they have collected, cleaned, and labelled — are building a durable competitive advantage. Companies that fine-tune foundation models with publicly available data are not. The data asset is the moat.
Does the AI create organisational knowledge? When employees use AI tools, does the organisation become measurably more capable over time? Is institutional knowledge being captured, structured, and compounded? Or is AI usage ad hoc and individual, creating no organisational memory?
Are there measurable network effects? Does the AI system improve as more users interact with it? Does customer usage generate data that makes the product better for all customers? Network effects in AI create the same self-reinforcing advantage they created for social platforms.
Is the AI capability defensible? Could a competitor replicate the AI capability in 12 months by subscribing to the same APIs? Or would replication require years of data collection, model training, and organisational transformation? Defensibility is the test of asset durability.
Does the company measure its intangible assets? This may be the most revealing question. Companies that understand they are building intangible assets measure them — they track data quality, model performance, organisational learning, customer relationship depth. Companies that are AI-washing do not measure because there is nothing to measure.
When the dot-com bubble burst, the companies that survived and thrived were not those with the most traffic or the highest valuations. They were the companies that had built genuine intangible asset portfolios: Amazon's logistics infrastructure and customer data, Google's search index and advertiser relationships, eBay's marketplace network effects. These assets were invisible during the bubble but determined the outcome after it.
The AI correction will follow the same logic. When sentiment shifts — and in every technology cycle, sentiment eventually shifts — capital will flow away from AI companies that cannot demonstrate measurable intangible asset value and toward those that can. The Opagio questionnaire and valuator exist precisely to enable this distinction: making the invisible visible so that capital can be allocated based on genuine asset value rather than narrative momentum.
Consider two AI companies in the same sector, each valued at similar multiples. Company A has spent three years building a proprietary dataset of 50 million labelled interactions, has fine-tuned models achieving measurably superior performance on domain-specific tasks, and has built an AI-augmented workflow that reduces customer acquisition costs by 40%. Company B uses third-party AI APIs, has no proprietary data, and generates comparable revenue through aggressive marketing. In a market correction, Company A's intangible asset portfolio protects its valuation. Company B has nothing to protect.
I have watched this pattern three times now — the dot-com surge, the social media repricing, and the current AI cycle. At IG Group, I was building intangible assets — technology platforms, mobile-first products, customer relationships — while the City questioned why the technology cost base was high. The answer became clear over time: those investments created the assets that drove the company from a valuation of roughly three hundred million pounds to nearly three billion. The market could not see the assets on the balance sheet, but they were real and they compounded.
Facebook's IPO taught me the same lesson from the outside. The market initially could not value what it could not see — and then spent a decade catching up to the reality of intangible asset value that had been there all along. The subsequent acquisitions of WhatsApp and Instagram, initially criticised as wild overpayments, demonstrated that some participants in Silicon Valley understood intangible asset value intuitively, even when formal valuation frameworks could not capture it.
This experience is why I built Opagio. The gap between what companies build and what accounting standards recognise is not a minor inefficiency — it is a structural blind spot that causes systematic mispricing. In the AI era, where the majority of value creation is intangible by definition, this blind spot is larger and more consequential than ever.
The AI bubble will correct. Some AI companies will prove to be genuinely transformative; others will prove to be the Pets.com of this generation. The difference will not be in the quality of their demos or the size of their funding rounds. It will be in their intangible asset portfolios — the proprietary data, the trained models, the organisational knowledge, the customer relationships. Opagio's role is to make these assets visible before the correction, not after it. Markets that can see intangible value can price it correctly — and investors who can measure intangible assets will be positioned on the right side of the repricing.
The practical implications are clear. Investors — whether PE firms, venture capitalists, or institutional allocators — need to integrate intangible asset assessment into their AI investment process. The AI-washing checklist provides the first filter: is the AI real? The intangible asset test provides the second: is the AI building durable value?
Three actions for the current market:
Audit the intangible asset portfolio. For every AI investment, map the intangible assets being created: proprietary data, trained models, organisational processes, customer relationships, brand capability. If the map is empty, the investment is fragile.
Measure the compounding rate. Genuine intangible assets compound — datasets grow, models improve, organisational knowledge deepens. Track whether the AI investment is generating assets that increase in value over time or whether it is a static deployment with no asset accumulation.
Stress-test the valuation without the AI narrative. Remove the AI premium from the valuation model. What is the company worth based on current revenue, margin, and growth trajectory alone? If the answer is materially lower, the gap represents the intangible asset value that must be verified — or the bubble premium that will eventually deflate.
Ivan Gowan is Founder and CEO of Opagio. He spent 15 years as a senior technology leader at IG Group (LSE: IGG), overseeing engineering growth from 4 to 250 during the company's rise from £300m to £2.7bn. He built IG's first online and mobile trading platforms, launched the world's first Apple Watch trading app, and holds an MSc from Edinburgh with neural networks research (2001). Learn more about the Opagio team.
AI makes data more valuable, and data makes AI more capable — creating a symbiotic value relationship that compounds over time. This article examines how to measure, value, and invest in the AI-data flywheel that drives the most defensible technology businesses.
Read more →Most generative AI business cases focus on cost reduction, but the larger opportunity is revenue growth through new products, enhanced customer experiences, and market expansion. This article provides a framework for measuring generative AI's revenue-side impact.
Read more →AI licensing fees represent only 20-30% of total AI adoption costs. The remaining 70-80% — data preparation, integration engineering, change management, and ongoing maintenance — are the hidden expenses that derail AI budgets and timelines.
Read more →Get the latest insights on intangible asset growth and productivity delivered to your inbox.
Book a free consultation to see how the Opagio Growth Platform can help your business.