GDP Is Dead: How AI Forces a Rethink of Growth, Politics, and Power

Remember Simon Kuznets. Mid-20^th century economist, immigrant, and empiricist. He believed that economic output could be measured in a standardized form using mathematics and statistics and he gave the world the tool to do it: national income accounting and Gross Domestic Product (GDP).

Born 1901 in what is today Ukraine, Kuznets emigrated to the United States in 1922 after the turmoil unleashed by the collapse of the Russian empire and the Bolshevik Revolution. He came of age during a time of depression and war and was fascinated by economic change and the big forces that shape it. In the 1930s-40s he pioneered systemic ways to measure economic production, income, and consumption across an entire economy, creating the concept of GDP and winning him the Nobel Prize in 1971.

A radical thinker, he was ever critical of the tool he had created, warning that it did not take into account inequality, unpaid work, and environmental damage, famously cautioning that “the welfare of a nation can scarcely be inferred by a measure of national income.”

Despite cautionary warnings by its creator, GDP has nonetheless become the scoreboard of modern capitalism.

Until perhaps now. The advent of the AI economy and its profound implications may in fact be the death knell of this flawed but seemingly universal statistic.

Firstly, the core assumptions that back GDP are very specific and human labour centred: People work, firms hire, wages fund demand, output is scarce, production is costly, and prices signal value. Human capital, and the productivity of it, is one of the binding constraints on growth.

In a modern AI economy the constraint on growth shifts away from labour to compute, energy, data, capital, and raw materials. In this new world, rare earths and energy matter more than office workers. As a result, economic returns will increasingly shift to owners of capital and natural energy resources away from labour. This has the risk of creating huge income inequalities as the old economic models that tie wage growth to economic growth start breaking down.

In this new paradigm, output no longer equals employment, productivity no longer equals income, and growth no longer equals welfare.

Secondly, demand side metrics such as GDP will start becoming more misleading. One example of a potential paradox is that AI agents may produce lots of output, very cheaply, on the surface looking like GDP is actually contracting (as it measures prices).

This will likely bring with it a shift to more supply focused measurements that take into account how much natural resources an economy has access to. As the framework shifts towards the quantity of energy and raw materials we can harvest, issues around environmental balance will become unavoidable.

On the one hand we will have a hungry machine that demand energy and natural resources at enormous scales, while on the other hand we will have the spectre of global warming and environmental imbalance. We will need to make choices between harvesting the “free” resources of the environment versus more economic growth, and the current framework of GDP is not equipped to properly account for debit and credit side of this trade off.

It is very likely that a new sort of politics that address wealth re-distribution and environmental factors will start to emerge across the world, and we are in fact seeing early examples of this today in many places. Likewise, a new geopolitics focused on energy, data, compute, and natural resources is already emerging.

What is ultimately at stake is not a better statistic, but a better understanding of power. GDP told us who was productive in a labour economy. AI demands that we understand who controls energy, compute, data, and capital in a post-labour one. Measurement shapes policy, and policy shapes outcomes. When the measure fails, politics fills the gap.

GDP was the right tool for an industrial century. The AI age will require something harder, more physical, and more honest, one that acknowledges that growth is no longer constrained by human effort alone, but by energy, resources, and the choices we make about how to deploy them.

How AI will Transform Fintech: Cross-border SMB Commerce Rails

Today the cross-border logistics plus payments stack is broken for merchants under $10-20 mn of revenue.

Much like sewing together a patchwork quilt from scratch, these merchants today have to stitch together Shopify, ShipBob, UPS, Avalara, Wise, local import brokers, customs paperwork, random tax agents, and then sit down and pray nothing gets stuck at some border.

Economically speaking, landed cost miscalculation is also painful: overcharging kills conversion for merchants, while undercharging eats up margins. Trying to do this yourself eats up valuable time.

Customs classifications are a black box, HS codes are arcane, and the wrong code once again costs time and money.

Returns are pure chaos, and even something as simple as understanding and optimizing FX spreads is daunting for merchants.

And if you are an SMB, bigger players like the shippers or Global-E don’t really pay too much attention to you.

An AI driven platform can change all this, and Swap Commerce, where we led the Series-A is already tackling this with great success.

The vision for Swap is clear:

A real time accurate landed cost calculation (including duties, VAT, brokerage fees, etc.) delivered in milliseconds during checkout
Customs and compliance automation with the correct import/export declarations and registrations handled smoothly and quickly
Payments and FX bundled, also offering local payments methods, even acting as the merchant’s cross-border PSP
Orchestrate all logistics including shipping labels, duties paid options, returns routing, and warehouse integrations
A returns clearinghouse that can offer in-country consolidation hubs and allow cheaper returns
Localizing pricing, currency and tax display on a country-by-country basis
Fraud and risk management by tracking cross-border specific false positives and false negatives a traditional PSP may miss.

The regulatory volatility today, starting from tariffs and stretching back to Brexit all have created an even more urgent need to solve this pain point. And AI is making compliance automation solvable in a way it had never been up to now.

So, we are understandably very excited about the prospects for Swap. As we continue to scale up, what appears to be a workflow tool also reveals crucial network effects. HS classification training sets, customs and return data, lane-level logistics performance data all give Swap an edge that gets even stronger as it scales.

We will look at other AI driven fintech opportunities in future blogs, and analyze some of the common themes that emerge from these opportunities, but by ways of a spoiler alert, Swap is an example of two very salient themes that are emerging in fintech AI:

Combining the power of AI with a deep understanding of regulation & compliance, and
Using AI, automation, payments rails etc. to present a bundled solution to the customer as opposed to just the individual components of software or automation or payments.

Another way to think of the above is that the fintech winners of tomorrow will combine AI & automation, an understanding of regulation & compliance, and embedded finance into one solution. We can think of this as Solution-as-a-Service, and the companies, like Swap, that deliver the best will be the winners.

In future blogs we will look at how these themes will transform other areas such as AI driven treasury ops, real-time compliance oversight at regulated entities, AI driven NPL solutions, and many more.

Best Practices in Fintech AI – Notes from our CEO conference

We recently had our 17^th Annual QED CEO Summit in Washington, D.C. and had an amazing turnout from what we (admittedly somewhat biasedly!) think are the most exciting entrepreneurs and builders in the world of global fintech.

Given the times we live in, it should not come as a surprise that we had lots of great discussions around AI and how startups and scaleups are using it across several functions. In the spirit of continuing this very interesting (and certainly fast evolving) conversation, we wanted to share some of the key takeaways from these sessions.

I. AI is being used by all, but to varying degrees

One not so surprising insight was that when we asked CEOs to rate how much they have integrated AI to their organization on a scale of one to ten, nobody thought they were even close to a ten. More interestingly however, as CEOs listened to the practices of other peers, some changed their own rating from high to low single digits. The twofold takeaway here is that one the variance across organizations is indeed wide, and two that cross fertilization and sharing notes with others brings much improved awareness. So keep sharing, and keep learning – fast!

II. Data & prompt engineering: Common frameworks are emerging on agent workflows

Several teams are converging on various prompt-engineering frameworks (e.g., a simple one being Role-Context-Task-Format). But consistency varies widely across companies, and we’re still in the early phase of codifying ‘how we work with agents’ and the norms are evolving very quickly.

This then leads to a key skill set around writing prompts, or in other words “managing agents” developing rapidly and becoming very valuable.

In this fast evolving world, it is also quite common to use one agent to write a prompt for another agent, for example using ChatGPT to write a prompt for Claude.

While there were many examples of note here and lots of best practices, things are changing so quickly, that the most salient point was the emergence of a new “AI culture” where speed of adaptation is becoming a defining characteristic.

III. Agent + human workflows: A new sort of team

Workflows are changing very fast also, and it is clear that all CEOs now think of their teams as being made up of humans plus agents. Most CEOs are finding ways to incorporate agents into their teams and workflows in creative ways, where each one supplements the other.

One interesting example that was shared in the financial infrastructure industry involved an AI agent joining a customer product discovery call, and then based on the notes from that call writing prompts for another AI agent to create new wireframes for the new product that was being discussed. According to this CEO, just using this simple process cut down on developer needs by a factor of five when measured in manhours over a period of two months, and also enabled one product manager to cover three times as many releases in one quarter.

IV. Specific issues around AI in fintech

In the context of fintech, one very important external interaction to manage was around regulators. Most fintech CEOs were aware that regulators would likely be slow to adapt, and there were very valid concerns the community would need to answer around model drift, fairness, explainability, and data lineage with regulators. One best practice that was brought up here was to share data and insights with regulators in an open manner, bringing them along on the journey.

V. HR impact is emerging as not everybody can keep pace

A concrete example is that while common frameworks on how to generate the most effective prompts most efficiently are quickly emerging, not everybody in the organization is able to keep up. As a result, there is certainly an issue around what to do with those in an organization that are not able to adapt to this new world at equal speed.

The not so surprising upshot of this is that organizations are parting ways with what we may refer to as AI luddites (not an ideal outcome for anyone to say the least), but also a lot of thought is going into how to upskill the workforce to adapt to this new world, and how to incentivize everybody to learn and adapt their mindset.

The importance of adapting the firm’s culture to AI becomes very important as a result, creating best practices around how AI is used, driving adaption, and finding ways to share insights quickly.

VI. A new type of tech debt: some words of caution

One interesting point was also made around AI generated product prototypes creating a new sort of tech debt or friction in the product discovery and design process. The issue here was that a simple product prototype created by an AI (as in the example above) may not necessarily be easy to productionize due to a host of technical issues. The best practice shared in this case was to treat the AI created demo as a “throwaway prototype” or a PoC and then start from scratch (or at least give the team latitude to make the needed layout changes) later in the development process.

VII. Managing interactions with the outside world:

What also became clear is that while managing agents and their interactions with humans was a new and emerging skill set, managing their interactions with the outside world was another.

One example of this was an interesting theme around B2B focused businesses finding ways to educate their business customers on how to use AI in implementation. While managing internal culture and adaption was one thing, managing the customer’s adaption was another altogether. An example brought up here by one CEO as that for each dollar spent on AI development, they were budgeting one dollar for helping their customers with adaption.

VIII. Sharing best practices is empowering

A key insight is that CEOs loved the sharing of insights enabled by our conference.

In this spirit of sharing, if you found this interesting also make sure to refer to the additional content put out by the QED team around this topic for much more, including signing up to our newsletters and this blog. And most importantly, as you continue building your fintech business, do reach out to our global team – we’d love to hear your thoughts and incorporate your perspectives, especially around how you are using AI to compete – and win.

Thoughts on a New Era: From Wheat to Networks

Human society is now rapidly evolving from a services-based economy to a network and intelligence-based economy.

But what does that mean, and what can we expect from the future? The following broad historical view could be helpful in putting current changes into context.

The hunter-gatherer stage of human evolution kicked off as early as 2.5 million years ago with early species such as Homo habilis (when tool use and foraging first emerged). In terms of modern-day humans, we could say that the start was with Homo sapiens around 300,000 years ago. The default mode of production was the use of simple tools to hunt and trap various animals and dig up tubers, crack open hard-shelled fruits etc.

This was the default mode of human existence until 12,000 years ago, when a big technological leap occurred: humans domesticated wheat and other plants. As an interesting side note, looking at Earth from space, an alien may argue it was actually wheat that domesticated humans – if you find this topic interesting a good book to read is Oceans of Grain.

The move to farming was a big leap indeed. As a result, human populations across the world increased from an estimate of 1 – 10 million to about 150 – 300 million by 1 CE. The farming revolution also allowed for excess food production that could now employ people not involved in the production of sustenance and food gathering. Thus armies, kings and queens, priests, and empires entered the rapidly growing world. In many ways, we can say that this was the beginning of history. Think the Mayans, Ancient Greece, and the Roman Empire.

The farming boom lasted about 11,000 years until in around 1760 CE the industrial revolution kicked off in Britain. It started with mechanization (think spinning jenny, the steam engine, etc.) and was fuelled by coal and urbanization. Human population once again turbo charged, going from 770 million globally in 1760 to 2.5 billion in 1950.

The industrial age lasted even shorter, in total less than two hundred years until the mid-20^th century when the services-based economy and the digital age was ushered in. This was characterized by the first computers, the growth in financial markets. There were big population and employment shifts. Just as workers had moved out of farms and into factories in the industrial age, those workers now moved into offices in the services age. Population also ballooned, to 8 billion in 2022.

And what were the factors of production in each of these ages? For hunter-gatherers it was wild resources and human effort with no capital beyond simple tools to speak of. Entrepreneurship just meant taking risk to survive in the literal jungle.

As we moved into farming, wild resources were replaced with cultivated fields, a labour class of farmers and herders emerged, complemented with specialized classes of priests, soldiers and administrators. Capital tools got an upgrade, animals were domesticated. Entrepreneurship started emerging with landlords, traders, and innovations like crop rotation.

In the industrial age, labour continued to specialize even further, with concentrated factories urban centres started growing, and as heavy investment such as factories and infrastructure were needed, capital entered the stage in a big way as a factor of production. It is no surprise that Das Kapital was written during this transformative age. Hand in hand with capital, entrepreneurship also evolved, risk was scaling and robber barons such as Rockefeller, Watt and J.P. Morgan entered history.

In the services age, knowledge workers that used their brains instead of muscles emerged. Capital shifted to include more abstract things such as brands and patents. We can also think of this as the knowledge economy, and human labour was still a limiting factor, but more because of their knowledge and reasoning capacity, not their arms and legs.

We have now entered the network and intelligence age, and the factors of production, as well as the limiting factors on economic growth are once again shifting. On the labour front, humans that used to be knowledge workers in offices are now rapidly moving out, being replaced by computers that can reason faster, more consistently and more accurately.

As a result, on the capital front, the main constraint of growth is shifting from humans to computational power (electricity plus advanced chips). Economic moats are not built around big offices or big factories, but proprietary networks and the data they generate.

The networks of this age are multi-faceted and intertwined. On top of communication networks (satellites and cables) we have financial networks (Visa, MasterCard, Swift), as well as social networks (X and others). These networks generate vast amounts of data that in turn fuels the vast computer intelligence that is evolving ever faster.

This age is also characterized by ever faster innovation and disruption, yet those that control the networks and the computers will yield unprecedented power.

So what does this mean for entrepreneurs working with QED and building in this age? Access to proprietary data and building a network is most certainly the holy grail. If your business does not have strong elements of this, even if in a niche form, you may want to reassess your business plan.

Given that the pace of innovation and disruption is increasing, opportunities for entrepreneurs are also multiplying. Look for incumbents that are hampered by regulation and may be slow to react to the new age.

The skills that are needed in this age are agility, speed, adaptability and calculated risk taking. Taken together, these amount to being anti-fragile – building organizations that emerge stronger from each successive disruption and shock. You will also have to be good at incorporating non-human agents into your org structure. Sounds simple, but laws, regulations, and human nature will complicate it.

Yet capital is still needed. Computing power will not be free, whether from humans or machines. And acquiring customers still costs money. As QED, we are here to help.

Who Will Win in AI: Data, Regulation, and Power

AI is continuing to transform our lives, and not a day goes by without a new announcement, a new investment, or a new model being released. But who will be the long term winners in this field?

If we look back to what was the dawn of the e-commerce era in the nineties, a similar pace of investment and innovation was taking place during the internet boom, and the ultimate winners were not easy to predict.

In December 1997 Amazon had a market cap of $800 million, and eBay had not yet gone public. By December 1998 Amazon was up more than 20x at $19 billion, and eBay, fresh off its IPO traded at $11 billion. As another year went by and the millennium drew to a close in 1999, Amazon traded at $270 billion vs $17 billion for eBay.

Then the dot com bubble burst. By December 2003, Amazon was trading at $11bn and eBay had surpassed it (driven by a more profit oriented performance) to $21bn. Roll forward to more than two decades later, as of April 2025, Amazon is now at $2 trillion vs. $32 billion for eBay.

If we look at other big technological innovation, for example smartphones, similar patterns emerge. In 2007, Nokia’s market cap was $117 billion, larger than that of Apple at $112 billion. Today, Apple is at about $3.3 trillion vs. $29 billion for Nokia, and very few people would have heard about the Symbian smartphone Nokia was launching in 2007. Another aspiring smartphone maker, Palm, traded at a peak valuation of $53 billion, but today many people have never seen a Palm Pilot, much less the Treo that Palm launched in 2003 as a smartphone.

So let’s switch back to AI. Who will be the dominant players of tomorrow? Given the above, it may not be easy to predict.

In simple terms, the inputs needed to win are vast amounts of data and power (both computational and electrical) along with distribution channels. This requirement already gives certain existing players a big advantage.

Tech giants such as Alphabet and X have access to vast amounts of proprietary data in the form of all our e-mails and social media content. With these existing user bases, distribution is also easier. Advantage incumbents in this case.

Yes, ChatGPT has a solid first mover advantage, but it will likely have to work harder to overcome the data edge of Gemini and Grok. Is this another eBay vs. Amazon moment?

Furthermore, the sophistication and scale of AI models are escalating rapidly, directly impacting who can compete. Today’s leading models—like OpenAI’s GPT-4o, Google’s Gemini 2.0, or xAI’s Grok 3—boast parameter counts in the hundreds of billions, requiring massive datasets and compute resources to train. For instance, training a model like GPT-3 (175 billion parameters) reportedly took 3.14 × 10²³ FLOPs (floating-point operations), a number that’s ballooned with successors. Again, advantage incumbents.

In the world of fintech, banking giants such as JPMorgan and Capital One also have access to decades worth of proprietary user data. Likewise, they have a distribution advantage, but they will have to fight the “compliance says no” mindset in bringing innovation forward, a story we are familiar with as fintech investors at QED. Advantage incumbents, as long as they don’t drown in their own red tape. Most certainly an exciting opening for aspiring challengers.

All these examples are from the United States, but looking through a geopolitical lens, we see the advantage China also has. In addition to the vast amounts of data generated by its population it is also the biggest solar panel producer globally which gives them an advantage on the power dimension – the marginal cost of solar power is close to zero (as low as $0.01/kWh). China currently produces about 80% of solar panels globally, plenty to power future GPU farms.

Europe, on the other hand, has been a laggard. Using regulation to try to even the playing field will be tempting, so we can expect an amplification of the debate around who owns user data and GDPR. This is an important debate to have, but not at the expense of stifling innovation. And there is strong impetus around European domiciled models, and many exciting startups and scaleups, so expect more in this space. Europe cannot afford to miss the boat on AI.

This also highlights another incumbent with access to both data and power: national governments and their various branches, including the armed services. As the importance of data and AI becomes clearer, there will be a temptation for many governments to collect and store more data on its citizens, and then use that data to power AI models. To ponder the possible social and political implications of this, the book Nexus by Yuval Noah Harari is a good starting point.

So while it looks like the incumbents look to lead the AI race with data and power, we are VCs and exist on the premise of disruption and the power of innovation. AI creates plenty of opportunity, especially in fintech where regulation is difficult to navigate for many.

The so what for fintech entrepreneurs building in the AI space is this:

First, find your own data edge and grow it and guard it jealously. If your business model gives you proprietary data, this is your doorway into a future AI moat. Use it wisely.

Second, regulation is more cumbersome for giants. Use regulatory arbitrage to your advantage while you can. Be nimble and take calculated risks. You need to understand regulation, yet not be hamstrung by it.

Thirdly, look to niche areas to grow from. Amazon started as a bookseller and became a global e-commerce giant. Fintech presents an interesting wedge for founders, where niche data sets can be sharpened to provide an edge to a targeted user base with domain specific models. Think fraud detection, the HR channel and the future of work, auto finance and insurance, tax and tariff complexities, and many others.

From there, be the best operator out there and out-maneuver your competition, adding one new vertical or market at a time. And QED is here, ready to help with our global reach and understanding of regulatory businesses.

Finally, look for where AI is disrupting incumbents. Regulation slowing down banks, making them laggards in deploying AI may be one example. Another area may be in distribution, and the emergence of voice as a distribution channel. We often say that the smart phone is a very clunky input/output channel with a small screen and small keyboard for our two thumbs. Being able to ditch your screen and talk to your AI, having it summarize your inbox, reply on your behalf etc. may supplement our use of smartphones. More on that in our next blog.