From Charlie Munger to ChatGPT - The 'productivity gains' paradox
If you're seeing productivity gains in your core business, you're losing already.
Munger tells this oft-quoted story about how profits move in a value chain on the introduction of a new technology:
When we were in the textile business, one day, the people came to Warren and said, “They’ve invented a new loom that we think will do twice as much work as our old ones.” And Warren said, “Gee, I hope this doesn’t work because if it does, I’m going to close the mill.”
He knew that the huge productivity increases that would come from a better machine introduced into the production of a commodity product would all go to the benefit of the buyers of the textiles. Nothing was going to stick to our ribs as owners.
And it isn’t that the machines weren’t better. It’s just that the savings didn’t go to you. The cost reductions came through all right. But the benefit of the cost reductions didn’t go to the guy who bought the equipment. It’s such a simple idea. It’s so basic. And yet it’s so often forgotten.
Yes! It’s so basic, And yet it’s so often forgotten.
This is a tale as old as technology itself. And , it’s playing out with Gen AI all over again.
If knowledge work becomes extremely cheap, and is one well-structured prompt away, it’s not too different from a new loom delivering 10x productivity gains.
And yet, the very businesses that get excited about short-term productivity gains are the ones losing long-term competitive advantage.
And that - is the productivity gains paradox!
LinkedIn is flooded with companies claiming productivity gains. If those productivity gains are in their core product value chain, they’re - in one word - screwed!
Who captures profits from the gains?
You’re in the restaurant business, running a wildly popular spot in a prime location. Your success draws crowds, but when it’s time to renew your lease, the landlord demands a massive rent increase. Suddenly, the very space that helped build your brand feels like a trap.
You have two options - agree to the higher rent, or buy out the land.
If you decide to do the latter, you’re now in the real estate business. You need to think through whether all the investments you’ve made into your current location, and whether its location advantage, justify the premium you pay for it. Or whether you’re better off finding an alternative cheaper location and rebuilding your brand there.
Something similar confronted Netflix when it went into streaming.
If it licensed all its shows, it was stuck being a distributor in a business where pricing power is held almost entirely by content creators. Capturing and retaining users is expensive and difficult. And yet, if Netflix found a way to do that, studios would simply increase licensing fees, preventing Netflix from holding onto profits.
Netflix had no choice but to create its own content.
Look at Spotify, on the other hand, which still struggles with that problem as labels squeeze out 70% of the revenues. I’ve talked previously about ways to counter that but there are limits to how effectively you can counter such power when you’re stuck in the middle.
This is playing out with AI right now. If you’re a knowledge services firm getting excited about the productivity gains that AI is delivering to your core business, be very concerned. Because long term pricing power sits with those providing AI.
ChatGPT jumping its Pro subscription price to $200 is only one very small manifestation of that.
Who gains scale advantages?
When assessing the impact of technological shifts, it’s critical to identify who gains scale advantages.
Consider the arrival of the internet and its promise of digital distribution.
News companies jumped to shed their costs - printing presses and proprietary distribution networks - and embraced the newfound productivity gains with an asset-light business model.
However, they soon realized that the very benefits of free distribution were universally accessible, enabling a flood of competitors to enter the market.
Ironically, the high fixed costs of printing presses and distribution networks had previously acted as barriers to entry, protecting incumbents. By eroding these fixed costs, the new technology drove up competition eroding their margins.
Moreover, the shift to digital introduced new variable costs—specialized digital media expertise—that were expensive and scarce, further straining the business model.
Compounding these challenges, the news product itself became unbundled, with content separating from advertising. This unbundling siphoned away profit pools that news companies once relied on, leaving them stuck with higher costs and diminished revenue streams.
The scale advantages shifted from news production and distribution to news curation and personalization.
The productivity gains paradox
A parallel can be drawn to Generative AI today. Companies often rush to adopt AI for productivity gains without realizing the unintended consequences.
Years of investment in deep expertise and knowledge - their traditional competitive moat - can quickly be deconstructed into variable costs.
By doing so, they risk ceding pricing power to others, as the barriers that once protected them are dismantled.
And by relying on a supplier of ‘knowledge’, they fall into the same trap that Netflix, Spotify, or your successful restaurant fall into.
That’s Charlie Munger’s paradox hitting you right there.
What works and what doesn’t
When are productivity gains valuable, and when are they early signs of impending doom.
Ask yourself three questions to figure this out:
Are productivity gains delivered on your core production chain or in complementary functions?
Klarna, a fintech company, highlights the productivity gains it has achieved by replacing its support staff with AI agents.
This illustrates a strategic use of productivity improvements in complementary functions rather than in the core value chain.
Enhancing customer support, for instance, can lead to a better customer experience, which in turn drives greater adoption and usage of the core product - payments in Klarna’s case. This is an example of how productivity gains in complements can reinforce and amplify the value of the main offering.
However, this dynamic reverses if productivity gains are applied to the core product itself. For example, if a call center - a business whose core product is delivering customer support - uses AI to replace its primary service, it risks undermining the very basis of its value proposition.
In such cases, you’re handing over critical value drivers in your core product to a supplier. You land in the classic Porter’s ‘bargaining power of suppliers’ trap.
This distinction is critical for knowledge-based service companies. They must ask themselves:
Are the productivity gains being achieved in complementary areas that enhance the core product’s adoption and value?
Or are they changing the economics of the core product itself?
Can you change the economics in your favour by shifting profit pools?
Even if you operate in an industry where the productivity gains of generative AI significantly impact your core product, can you still control the economics in your favor?
Often, the most effective strategy is to find ways to shift profits toward complementary capabilities.
By enhancing and monetizing these complements, you can maintain pricing power and profitability, even as the core product becomes subject to supplier pricing power.
Spotify, as we saw above, is a classic case of the problem with being held hostage by supplier pricing power. The music labels capture 70%, while Spotify does all the grunt work of acquiring and retaining customers.
In contrast, the iPod business model counterbalanced the music labels’ leverage by shifting its profits away from the music value chain and making money on a complementary device instead of the music. Had Apple done the opposite —subsidizing devices while monetizing music —it would have landed in the same problem.
Can you leverage AI to create proprietary production advantages?
It’s still possible to enhance profits and secure a stronger competitive advantage with generative AI in your core product- but only if you leverage it to build a proprietary product advantage within your business. The key is ensuring that this advantage is developed internally, rather than relying on a third-party supplier of AI.
Relying on third-party AI providers without strong proprietary advantages or knowledge leaves you vulnerable to their pricing power. Additionally, the entry barriers to your industry could diminish, as others gain access to the same AI tools and capabilities, increasing competition and reducing your differentiation.
Quick test to figure if you’re screwed…
If you believe AI won’t replace your business and that some ‘secret sauce’ or ‘human element’ will make you stronger in the age of AI, you need to be clear about how that advantage works.
Remember how media companies faltered when the internet introduced cheaper distribution?
That same technology shift transformed everyone into a media company. FMCG companies, for instance, leveraged content to establish direct channels to customers, effectively entering the media space.
They thrived in the business of complements - selling consumer goods - through content and direct engagement. The media companies, meanwhile, still struggle.
Either you’re
(1) in a business of complements, or
(2) shifting profits away from the core product you used to sell, or
(3) building proprietary advantages internally rather than relying on vague notions of human superiority augmented by AI.
If you can’t articulate your advantage in one of these three ways, you likely don’t have one.
Final note
I’ve previously written about how you can leverage all 3 effects in your favour in this piece on How to win at Generative AI. The four-step approach is written in the context of AI startups but can also be applied to knowledge services firms.
Thoroughly enjoyed reading it! The problem with not adopting AI in core capability is if we don’t do it competitors will. So there is no way to stop it.
It is quite ironic in a way that right now the big 4 are making the most amount of money in Gen AI, how long they can continue with that model or hold into the promise of how Gen AI will help their client in the near future.
I guess we need to see for another one or two years to see whether they can spin in a different way