The Coca-Cola’s of AI

(as well as its Walmart’s, Instagram’s, and its picks n’ shovels)

TL;DR;

Like other waves of innovations before it, AI is expected to create enormous opportunities both for companies that will provide the core technology needed to implement this new new thing, as well as for companies that are prescient in adopting AI to truly redefine their businesses. In the past, the most disruptive tech waves not only begat new era-defining businesses but helped set the foundations for whole new industries to be born. The impact of AI is expected to follow a similar trajectory.


“The people who invented refrigeration made some money, but most of the money was made by Coca-Cola which used refrigeration to build an empire… I view these large language models as refrigeration. Will there be some money made there? I think so… but the Coca-Cola has yet to be built.”
Chamath Palihapitiya, Social Capital, ex-Facebook VP of Growth

As venture capitalists strive to find the best AI investment opportunities, they are mapping out the investment landscape, often splitting the companies into two main categories: First, companies developing core technology (as well as the providers of the infrastructure that is needed by the core technology), and second companies building applications of this core technology. As such, the historical analogy of refrigeration (the core technology layer) and Coca-Cola (the application layer) may be worth digging into, not least because refrigeration was one of those transformative technologies that had a steep adoption curve thereby radically changing our lives and enabling the creation and growth of so many other industries on top of its core technology.

Stop for a minute and think how our lives would have been different without refrigeration.


The Core Technology Layer – The Refrigerators of AI

 The first self-contained home refrigerator was invented in 1916 and became the Frigidaire company. Soon after, in 1922, two Swedish inventors founded Electrolux to market a new generation of fridges with better technology.  Five years later, the electric-appliance incumbent General Electric (GE) got into the business by introducing a mass market model. The market flourished, and within two decades refrigeration had penetrated 70% of US households with Electrolux and GE ultimately leading the market.

Today, venture investors are looking for the Electrolux’s of AI - core technology providers that will drive the AI wave. Such companies include: (i) Large Language Model (LLM) providers like OpenAI (ChatGPT), Google (Gemini), META (Llama) or Mistral; (ii) high throughput semiconductor companies – a space currently dominated by NVIDIA; and (iii) a plethora of Infrastructure software companies that are needed to create, customize, deploy and secure the LLMs and their usage. The early incumbent in the infrastructure category is Scale AI, a company that services all LLM providers, helping them collect, clean, and curate the data that gets ingested as well as to manage the process of refining the models so that they perform. Scale AI is thus the quintessential picks n’ shovels provider to participants in the AI wave. As AI hits prime time and addresses large swaths of the economy, one would expect the core technology providers above (that is OpenAI, NVIDIA and Scale AI) to benefit massively. While venture capitalists are betting on the fact that each of the above core technology segments will represent vast markets large enough to support many multi-billion-dollar companies, there is every indication several of the early incumbents will maintain leadership and thrive within this hugely expansionary market. Unlike their predecessor, these Refrigerators of AI are not just selling their technology one-off but rather have turned it into a platform with resilient subscription business models, aimed to thrive beyond the current gold rush and to withstand the inevitable cycles of the market: NVIDIA is not just providing chips, but packaging them into hard-to-replace integrated hardware and software analytics platforms; Open AI is not providing LLM models, but a consumer application (chat) as well as subscriptions to APIs that are being integrated into developers’ sticky work streams;  Scale AI is not just providing a human feedback loop to enhance model performance but a complete software platform to bring end-to-end data management best practices to AI – think ‘pick n’ shovels as a service’ perhaps. (OpenAI was recently valued at $80B, NVIDIA at $3Tr, and Scale AI at $14B).

“Today, venture investors are looking for the Electrolux’s of AI - core technology providers that will drive the AI wave.”


The Application Layer – The Coca-Cola’s of AI

Back in the early 1920’s, Coca-Cola was making most of its money from selling concentrate to soda fountains that were sold in soda bars and pharmacies. In fact, the Coca-Cola Company had already gone public with a $40M market cap in 1919, only to see its share price halve in 1920 due to concerns around raw materials. You could also imagine investors dumping the stock because they knew that soda fountains were old-tech, ready to be disrupted, and fearing that Coca-Cola didn’t yet have a strategy for the next big thing: Refrigeration. The company must have heard that message because it quickly pivoted its business: It aggressively expanded its bottling operations so that Coke bottles could be bought in stores and kept in people’s newly acquired fridges;  and it invested heavily in brand marketing so consumers would recognize Coke when they shopped. As a result, the value of the Company continued to grow: by 1927, the Company was 10x its IPO price; shares withstood the 1929 crash; and, over the next 20 years, as refrigeration proliferated, they saw another 10x increase.  Today the Coca-Cola Company is worth $270B, 100 times Electrolux! The value of its brand alone is estimated at ~$100B.

This same pattern has repeated multiple times in history. A more recent, eerily similar example is that of Facebook, a company that went public in 2012 as the dominant social network (Application) for desktop computers (Infrastructure) at a $100B valuation. Its stock price also suffered in the first year or so after its IPO, partly because its early forays into mobile apps had failed to gain traction. Investors worried that Facebook was stuck with old-tech (desktop PCs) and didn’t have a strategy for the next big thing – mobile smartphones. In 2012, only 11% of Facebook’s revenues came from mobile.  Like Coca-Cola, Facebook was able to re-orient its strategy and by 2018, >90% of its revenues came from mobile. Its market cap today is $1.2Tr.

Similarly, for some time now, leading companies have been figuring out how they can leverage the current next big thing – Artificial Intelligence – to transform their businesses. In fact, we believe (contrary to Chamath’s proposition) that many Coca-Cola’s of AI have already been created – these are the ones selling syrup to soda fountains while successfully transforming their business to take advantage of the next big thing. No doubt many will fail in this transformation: the tech analyst Ben Evans likes to point out that large incumbents like to add new technology as a feature without changing their core business processes - the equivalent of Coca-Cola putting a fridge in the CEO’s board room to show that it gets refrigeration, but not changing its bottling strategy. At the same time, there is no doubt AI will allow multiple Coca-Cola’s to emerge – companies that are agile enough to see how AI will transform their business and take advantage of the opportunity at hand.

“At the same time, there is no doubt AI will allow multiple Coca-Cola’s to emerge – companies that are agile enough to see how AI will transform their business and take advantage of the opportunity at hand. ”

We would expect that companies that have the greatest chance of making such a successful transition are ones that have multiple AI tailwinds expanding their market opportunity. (This was the case for Coca-Cola for example, which benefited from the refrigeration, as well as the automobile-ownership tailwinds). An example of a company that will be experiencing multiple AI tailwinds is Figma, the collaborative design startup that Adobe tried to buy for $20B. Both Adobe and Figma are in the design market, and both are applying AI as a feature, allowing users to design with natural language for instance. But beyond adding a feature, AI is expected to radically change the market that Figma plays in – that of bringing designers and programmers to collaborate on creating products. Indeed, AI has already changed programming, as more and more software programmers use AI to help them code every day, reporting >50% productivity gains from AI copilots. (Microsoft’s GitHub already has >1.8M paying copilot subscribers.). Over the coming years, AI is expected to make writing basic software dramatically easier, expanding the market for software creation, and increasing the need for that software to be coupled with Figma’s collaborative design platform. AI will not only enable established designers to up their game and produce higher quality designs, but it will also make it easier for non-designers to produce great design, again expanding the market for Figma’s product. Thus, AI will enable Figma to enhance its product and expand the addressable market for product design, but it will also create an enormous external tailwind in the form of an expanding market for AI-enabled coders who will need more design capabilities.

Figma is but one example of the companies poised to become the Coca-Cola’s of AI, using AI in their Applications to dominate their industries. No doubt, some of these companies will buy or merge with promising AI-native startups that are bringing exciting new AI-driven capabilities to their markets. Figma itself recently bought a company that was already working on an AI-powered natural language interface to Figma’s product. Going back to the Facebook example, it is important to remember that Facebook only managed to dominate mobile and turn its fortunes around after buying Instagram and WhatsApp, two venture-backed, mobile-native startups. Thus, the question is not only what the Coca-Cola’s and Facebook’s of AI will be, but which company will be its Instagram and WhatsApp, i.e., AI-native startups that exit for large valuations to an incumbent to exploit their distribution.

It is no wonder that venture capitalists and technologists believe the coming decade will be one of the most exciting and productive years the venture industry has ever seen. VCs are already seeing a plethora of AI-native startups, each planning to build new applications that will transform existing industry segments – from healthcare and finance to programming and manufacturing and beyond. A golden age of VC is in the cards.

“It is no wonder that venture capitalists and technologists believe the coming decade will be one of the most exciting and productive years the venture industry has ever seen. VCs are already seeing a plethora of AI-native startups, each planning to build new applications that will transform existing industry segments ”



New Industries – The Walmarts of AI

Before we end though, indulge us with one last historical analogy from the 1920’s when Michael Cullen, a manager at a small chain of grocery stores wrote to his bosses with a proposal to create “monstrous stores…located one to three blocks off the high rent district with plenty of parking space, and some to be operated as a semi-self-service store.” Cullen’s bosses didn’t even answer him. So, he quit his job and launched the first ever supermarket in 1930 on Long Island, completely re-imagining the retail concept. The supermarket, called King Kullen, was moderately successful and it still exists. More importantly, it propelled the creation of a new multi-billion industry.

Cullen’s intuition to build large stores outside of high rent districts, with parking lots, must have been motivated by the huge increase in car ownership he was witnessing. In fact car ownership in the US went from close to 0 in 1910 to around 60% of households when Cullen quit his job in the late 1920’s. People could now drive to do their shopping, thus it made sense to build scale in lower rent districts. That, however, was not the only secular trend that determined the success of supermarkets. Refrigerators also meant that supermarkets and consumers could keep fresh foods for longer, and over the next decades the food industry would also industrialize, as the likes of Coca-Cola scaled their production and distribution, and built brands that could be recognized by shoppers visiting these ever larger stores. Over time, these trends formed a self-re-enforcing cycle which transformed the multi-trillion-dollar food industry.

One can imagine that at the time, many grocery shops saw the expansion of refrigeration and thought ‘this is great, we get this; we’ll put a fridge in our shop too’. Or perhaps there must have been farmers who saw it and thought: ‘this is great: we can sell more meat to people who can now store them longer in their fridges.’ Both would have been technically right: the farmer and grocery shop owner would have certainly seen their business grow in the short term by adding a fridge feature to their business, but they would have been completely wrong in that they were missing the larger transformation brought on by the supermarket.

It would take another half century for Walmart to launch its nationwide expansion and dominate the category and eventually rise to dominate the top spot of the Fortune 500 for decades. Perhaps though, if the venture industry existed in the 1920’s, Michael Cullen would have approached a VC and gotten funding to expand much more rapidly, gaining a foothold not only in Long Island but throughout the US and the world. Though such a scenario is obviously highly speculative, we mention this thought experiment to outline what the role of Venture Capital would have been: To help Cullen open his first store (seed & early stage VC) and once proven successful, to recognize those secular trends and the impact they will have, and thus provide the capital and the tactical insights to help him quickly expand his stores and gain market dominance rapidly (growth stage VC), aiming to own this new industry worldwide.

Today AI is expected to create many of its own Coca-Cola’s, but if history is any guide, the greatest transformations to our lives may come in areas that we cannot yet easily anticipate, and which will only seem obvious in retrospect. History teaches us humility in judging Cullen’s managers for not seeing the potential for supermarkets…  or the consultants at McKinsey for predicting in 1980 that the total market for cell phones would be less than a million subscribers by 2000. When the genius Edison invented the phonograph to record sound, he thought the best use case was for office dictation – as it was obvious to everyone including him that music could only be listened to in concert halls, and not as a recording. Yet, by bringing on massive cost reductions while increasing quality over time, truly disruptive technologies create adjacent opportunities which are indeed hard for the smartest among us to recognize from the outset. Industries which never used that particular technology start to rethink the scope and scale of their business with the use of this new low-cost tool, setting about tectonic shifts which result in the emergence of enormous new industries, each with multiple era-defining corporate giants: those will be the supermarkets and Walmarts of AI.

“Today AI is expected to create many of its own Coca-Cola’s, but if history is any guide, the greatest transformations to our lives may come in areas that we cannot yet easily anticipate, and which will only seem obvious in retrospect”


Disclaimer

The content of this article has been approved and issued by TOP Venture SA for information purposes only and does not purport to be full or complete. The information and opinions contained in this document are for background information and discussion purposes only and do not purport to be full or complete. No information in this document should be construed as providing financial, investment or other professional advice.

AnalysisSalman Farmanfarmaian