How AI Will Help SMEs Kill the Conglomerate, and Why That's a Good Thing

It's strange to think of a time when human beings were not surrounded by the constant badgering of corporations.

But the endless drum of advertising is really quite recent.

It kicked into gear in the late nineteenth and early twentieth centuries, as industrial production began to outstrip what could be sold through simple local demand. Once firms could make more than could be immediately consumed, competition shifted. It was no longer enough to manufacture well, companies had to persuade, differentiate and occupy space in the consumer's mind.

Branding stopped being a mark on a crate that guaranteed quality. Instead, it has become something more: in our modern world making a purchase decision reflects more than mere necessity, it reflects our attempt to associate ourselves with traits we want to identify with. Why purchase an iPhone over a Samsung? Not because one makes better calls than the other, but because each carries a different set of associations about taste, status, and belonging. The modern brand does not just try to convince us their product is good. It tries to make us feel a sense of meaning by being associated with the qualities the brand portrays. In the words of Airbnb's former Head of Community Douglas Atkin: nowadays, we don't pray for meaning, we pay for it.

But corporations long predate modern advertising. Their political and economic force has always been extraordinary.

The roots of the modern Western business corporation date all the way back to medieval merchant partnerships. One of the earliest precursors was the Italian commenda, used in maritime trade by at least the eleventh century. In the commenda one party supplied capital, another travelled and traded, and the contract set out how gains and losses would be shared. This was one of the first examples of risk mitigation, and attempts have continuously been made since to enable merchants, financiers, and rulers to enjoy profitable upsides and avoid risky downsides.

Over the following centuries the commenda evolved into more complex tools: insurance, partnerships, joint ventures and bespoke contractual arrangements. While this worked to an extent, there still always remained a risk the adventurers would personally incur the losses of the expedition. A single voyage could make a fortune, but it could just as easily destroy one. Yet for a long time, there was no clean, durable form that separated the venture from the merchant.

Yet as overseas trade and colonial extraction became more profitable, the tools of reducing risk continued to advance. In an attempt to encourage more ventures, rulers (encouraged by merchants and financiers, and enticed by the prospect of sharing in the profits of new ventures) invented "charters" that could be granted to trading companies. These charters kickstarted the process of creating durable legal entities with privileges that outlived any single expedition and would protect those holding shares in the venture from liability for losses. Importantly, these grants often came bundled with monopoly rights.

It turns out the combination of legal personality, continuity, capital pooling, and monopoly privilege, is very attractive.

Investors were finally able to take part in high-risk ventures without undertaking personal liability, and it allowed states to outsource commercial expansion while sharing in the spoils.

It also invited abuse.

Once monopoly grants proved profitable, they spread like wildfire well beyond risky foreign trade. In England, the Crown used them to reward allies, raise revenue and entrench patronage. Grantees could restrict supply, raise prices and extract rents from domestic markets with little public benefit in return. The backlash was severe enough that Parliament passed the Statute of Monopolies in 1624, limiting the Crown's power to grant most new monopolies.

However, somewhat unsurprisingly, that did not end the corporate appetite for monopolistic control. It merely encouraged corporations to make the ambition less explicit. Ever since, one of the unspoken ambitions of the corporation has been to become unavoidable: to own the route, the customer, the supplier, the channel, or at least enough of each that competitors, and customers, have to live on its terms.

At its peak, the Standard Oil company controlled over 90% of the United States' oil refining capacity and exercised extreme control over pipelines and distribution, leading to its ultimate breakup through government intervention in 1911. Standard Oil had become a machine for dominating an industry through acquisitions, control over refining, leverage over transport and overwhelming scale.

Standard Oil had not been granted monopoly rights to oil, but had been incredibly effective at consolidating the market so that the power it exercised was a monopoly nonetheless. Yet, the company had, at least to an extent, tried to learn from the history of monopolies. They had left 10% of the market not under their control in part in an attempt to avoid accusations of being a monopoly, and they would often shift resources between different states to adjust their presence in any particular market depending on whether they wanted to destroy or encourage local competition. ¹

The conglomerate is the modern descendant of that logic. Though it may not always monopolise a single market in the old sense, often it does something subtler. It spreads across multiple sectors, internalises more functions, buys optionality, rolls up fragmented competitors and makes itself hard to challenge because it can cross-subsidise, outlast and out-distribute smaller firms. It becomes less a business and more a governing structure.

Late twentieth-century globalisation supercharged this model. The combination of cheaper logistics, trade liberalisation, financial engineering and deeper capital markets made it easier to manage sprawling operations across borders. Private equity and roll-up strategies came forth as aggregation mechanisms taking advantage of these opportunities, implementing a financial playbook that has become a modus operandi in pretty much every major Western economy today.

Not every large firm has become a literal monopoly though, and they don't want to be in order to avoid anti-trust regulators. However, many have become hard to confront because local competitors are no longer competing with the shop across town. They're competing with organisations financed to endure thinner margins for longer while centralising more and more administrative capability. A modern conglomerate doesn't operate power as explicitly as a corporation like Standard Oil did in the 20th century, or the British East India Company in the 18th century, but they are equally as parasitic as they arbitrarily control the competition within their markets.

But AI can even the playing field.

AI has triggered an interesting development in major organisations. The advent of AI is forcing companies to reexamine their operating models to determine 1) does a traditional organisational structure make sense in the age of AI? And 2) as more AI products are released are there still meaningful advantages to having large white-collar workforces?

For example, Jack Dorsey recently reduced the workforce of Block by 40% - over 4000 jobs. The reason? In Jack's own words:

"we're not making this decision because we're in trouble. our business is strong. gross profit continues to grow, we continue to serve more and more customers, and profitability is improving. but something has changed. we're already seeing that the intelligence tools we're creating and using, paired with smaller and flatter teams, are enabling a new way of working which fundamentally changes what it means to build and run a company. and that's accelerating rapidly."

And Jack's not alone in making these decisions or having these thoughts. Pretty much everyone in business today knows AI is, or will, make a huge impact on the way they operate. Many are overwhelmed, many are scared of being left behind, and a smaller few are excited about the future. It's unclear whether these changes will ultimately be positive or negative, but what is clear is that many of the traditional advantages of size will go by the way side.

I believe this presents an opportunity to reintroduce competition to economies and put power back into the hands of SMEs.

Now first, to be clear, AI does not remove every large-firm advantage, especially in certain sectors. Companies requiring large amounts of capital (mine sites, major industrial or construction projects, highly regulated sectors) will still result in large organisations and SMEs will remain largely uncompetitive.

But it does attack one factor that often matters most and most quietly: coordination and enterprise intelligence.

The question is not whether AI makes every SME as powerful as a global corporation. It clearly does not. The question is whether a great many of the functions that once required a giant corporate shell now require only software, good workflows and a small number of capable operators.

In many sectors, the answer is yes.

Most SME value from AI will not come from training proprietary models. It will come from harnessing frontier models that already exist and wrapping them in useful systems. Accounting workflows, customer service, legal triage, compliance monitoring, sales ops, forecasting, research, procurement, internal knowledge management, creative production and marketing analysis are all becoming cheaper to perform well. The expertise is increasingly rentable. Intelligence is increasingly on tap.

This matters because conglomerates were, in large part, management structures for handling business friction. If that friction falls sharply, then the justification for maintaining huge layers of bureaucracy weakens with it. The giant org chart stops being a source of strength and starts being a drag coefficient.

An AI-native SME can adopt tools function by function. It can replace weak operators, automate routine work, and reconfigure its workflows without asking permission from five layers of management. A conglomerate can also adopt AI, of course, but it has to do so across legacy systems, internal turf wars, compliance committees, budgeting cycles and political hierarchies. In many cases, as Jack can attest, its own size becomes the thing it must destroy.

This is why the likely effect of AI is not that every small business becomes huge. It is that fewer businesses need to become huge in order to be competent, and the nimbleness of small organisations is exactly what will provide a competitive advantage for small businesses over conglomerates.

But giving SMEs conglomerate level organisation and enterprise-grade intelligence is just the first layer - the first traditional conglomerate advantage that AI is destroying.

The argument gets even more interesting when it reaches brand and distribution.

It's easy to say that AI will commoditise execution. That part is almost obvious. The deeper question is whether it also commoditises the traditional machinery by which firms captured attention. Because if execution becomes cheap, distribution becomes key.

Traditionally one of the key advantages of being a conglomerate is having your brand be internationally recognised. This advantage is created overtime through repeat customer interactions with the conglomerate's communications or through third parties introducing the brand, such as family and friends.

In this space conglomerates have had natural advantages over SMEs: extensive marketing assets (e.g.: email lists, promotional partner relationships), local knowledge in international markets, budget for repeated marketing exposure or premium ad placements, and the economies of content production (the same ad campaign costs the same for a large brand as for an SME but by its nature the conglomerate gets more extensive use from the materials). But much of these capabilities are already becoming commoditised.

If everyone can generate competent copy, images, video variations, landing pages, targeting logic and testing plans, and distribute them en masse to a global audience immediately, then content abundance destroys its own scarcity. When messages become cheap, the channels they flood become noisier and less effective. That means a scarce asset shifts. If AI makes content cheap, cut-through becomes expensive.

For decades, large corporations enjoyed an advantage because they could broadcast and they could afford repetition. They could put the same message in front of everyone, everywhere, often enough that it became familiar. But in a saturated AI market, familiarity may no longer flow as easily from brute-force media spend. It may flow from proximity, participation and belonging.

Sometimes that proximity is geographic. For many services, local presence still matters. A nearby accountant, cafe, fitness studio, dentist, builder or community organiser has an inherent advantage if the surrounding market values repeated contact and trust. But the new local is not always physical. The internet has connected people to the world while isolating them from their neighbour, and in that environment an SME may be "local" to a niche interest rather than a suburb. A small business can serve a globally distributed tribe with a shared taste, aesthetic, hobby, politics or identity. A small business is not seeking to position itself as attracting a broad market in the same way a conglomerate would. This leads to the rise of specialised products and services based on niche interests.

An interesting comparison is the change in Youtube viewership over the years. Over the previous decades Youtube's total watched hours has continued to explode, but interestingly, average viewers per channel has decreased. As more and more creators enter the space, viewers, with the help of Youtube's algorithms, have been able to engage with content suiting their niche interests. So total viewership increased but average creator view count decreased. It draws an interesting case study when considering potential broader economic trends.

This matters because, as we said earlier, brand is no longer just a proxy for quality. In many consumer markets, it is an identity signal. People buy what helps them describe themselves to others and makes them feel "more like themself". This could imply small businesses could be better suited to tailoring their offerings to nicher interestes, but that fact could also cut both ways.

On one side, AI is eroding the cost advantage of producing polished branding assets and managing sophisticated campaigns. If everyone can look professional, then the differentiator shifts from polish to coherence. Smaller firms may be better at building a scene, a ritual, a recurring gathering or a genuine sense of membership than a sprawling corporation ever could be.

On the other side, the change could favour celebrity, trend and mass attention. Global icons and highly visible platforms can still aggregate desire at enormous scale. If consumers want to buy what is popular, then AI could strengthen the new cultural conglomerates: celebrity brands, platform-native labels and algorithmic tastemakers. Perhaps we will see increases on both sides.

This is why in-person events, memberships and community-led acquisition are so interesting. They are responses to digital saturation. A conglomerate can buy reach, but it is harder for it to manufacture intimacy. It can sponsor a conference, but it is much worse at hosting a scene.

None of this means the corporation disappears or that every industry becomes friendly to small firms.

There are at least three broad classes of business where size will remain structurally important.

The first is high-capex industry. Mining, energy, semiconductor fabrication, shipping, large-scale construction, aviation, utilities and other asset-heavy sectors cannot be willed into decentralisation by better software. Their economics are tied to vast fixed investment, political relationships and long planning cycles.

The second is foundational networks and regulated balance-sheet businesses. Social networks, major exchanges, payment systems, cloud infrastructure, core telecoms and parts of banking still benefit from trust concentration, licensing, reserve requirements, network effects and systemic importance. AI may make them more efficient, but it does not necessarily make them contestable.

The third is frontier research protected by capital intensity, regulation or deep technical barriers. Advanced biotech, aerospace, defence, neural interfaces and frontier AI labs are not likely to become cottage industries.

So the claim is not that SMEs will defeat large firms everywhere, it's that AI reduces the number of sectors in which bigness is inherently superior, and it weakens the internal logic of conglomeration in a much larger share of the economy than most people expect.

Why is this a good thing?

If this shift happens, it's good not because small businesses are morally pure. They're just people, so they're not. It's good because a more SME-heavy economy is usually more contestable, more plural and less bureaucratically dead.

Conglomerates are often good at extracting value from distance. They standardise, centralise and optimise for the median. That can produce efficiencies, but it also produces internal politics, slow adaptation and a tendency to flatten difference. The bigger the organisation, the more energy must be spent on preserving the organisation itself.

SMEs, by contrast, can be narrower, more legible and more accountable to a real customer base. They can specialise. They can reflect local knowledge, whether that locality is physical or cultural. They can fail without creating systemic shock. And if AI lowers the cost of competence, then more of them can survive on actual quality rather than being crushed by administrative overhead.

That means more room for real competition. More room for businesses that know their market intimately. More room for communities and experiences that are not optimised for the lowest-common-denominator consumer. More room, too, for profits and decision-making to be distributed across many firms rather than pooled inside a few immense managerial pyramids. It also means fewer cases where an entire region, profession or supplier network is effectively hostage to the decisions of a handful of giant employers or buyers.

It also means less need to treat every human problem as one best solved by a giant central institution. In many domains, the twentieth century was an age of organisational giants because coordination was hard. If coordination becomes cheap, there is no reason to keep pretending that giants are efficient by default.

There is, of course, a darker path.

The bad version is not that SMEs lose. It is that everybody loses except the owners of the most foundational platforms, models, compute and political chokepoints. In that world, automation strips ordinary people of bargaining power, software abundance crushes margins everywhere else, and society drifts toward a kind of techno-feudalism: nominally open markets sitting on top of a small number of inescapable infrastructures.

That is a real possibility. There is no guarantee that AI decentralises by default. Much depends on whether the surrounding legal, financial and technical systems remain open enough for small firms to adopt intelligence without becoming dependent serfs of a few providers.

But if AI does what it appears capable of doing - reducing business friction, commoditising large parts of administration, and making both competence and infrastructure cheaper to access - then the centre of gravity should move away from the conglomerate and back toward smaller, more adaptive firms.

That would not make humans obsolete. It would change the unit through which humans organise productive life.

The corporation was one answer to the problem of risk, distance and coordination. The conglomerate was its twentieth-century hypertrophy. AI may not kill the corporation, but it may finally kill the idea that bigger is naturally better, and that would be one of the healthier economic shifts of the century.