agog

Hyper-scale; The Ambition and Excess of Big Tech’s Data Empires, Paris Marx 2026

Amazon completed seven data centers in St. Joseph County, Indiana, alone in 2025, each covering about 220,000 square feet, with plans to add up to thirty-two buildings on on site. Open AI brought two nearly 500,000 square foot facilities online in Abilene, Texas, planning another six data centers alongside them by mid 2026. That entire facility would eventually hold more than 450,000 graphics processing units (GPUs)–the hardware most prized for AI workloads.

Elon Musk brought a 785,000 square foot factory in Memphis, Tennessee, and quickly filled it with 200,000 GPUs, then bought an even larger facility that would house half an million of them…To top off the exuberance, the Meta CEO (Zuckerberg) bought up more than 3,650 acres of land in Louisiana–the equivalent of about 2,700 football fields–where he plans to build an AI complex so large the company could  power five million homes with the ten power plants it’s constructing nearby.

There are between ten thousand and twelve thousand data centers around the world. Only 1,300 of them were estimated to be hyper-scale by the end of 2025, but they were responsible for about 45 percent of worldwide data center capacity–a figure expected to rise to 60 percent by 2030.

Phoenix AZ has about 200 data centers with 25 hyper-scale data centers operating or planned.

They (hyper-scale data centers) are in a category of their own, and that means they need to be treated that way. The giants of Silicon Valley have been continually making the digital services we depend on more computationally intensive for years. Consider the push for streaming media, the effort to sell more data-dependent tools to businesses, or even the way so may of the websites we visit are crammed with ad trackers. But their ambition to add advanced AI tools into every facet of our lives has necessitated a massive, global build out of hyper-scale data centers costing them hundreds of billions of dollars with every passing year.

Those data centers don’t just need a lot more computer hardware to run them; they also need the energy to power them and the water to cool them. For many communities those demands are getting far too high — but the companies want to push ahead regardless. People are not just getting disillusioned,; they’re getting angry. And they have every right to be.

Campaigns to stop data centers have grown in cities and towns across many countries. Some have even succeeded in pushing the companies out of their neighborhoods. As the demands have grown, so too has the political pressure to do something about it. Places like Ireland, the Netherlands, and even parts of the United States have placed moratoriums on new data centers, and candidates have started running with the intention to expand them (moratoriums).

Amazon, Microsoft, and Google alone control 59 percent of the hyper-scale data centers around the world.

For the past several decades, tech companies have pushed a particular vision of digital technology; one defined by vast data collection that has spawned deep concern about privacy and surveillance. It’s impossible to disentangle the infrastructure from all the harm it has enabled. People are not only starting to take on data centers; they’re also starting to ask whether we need all this computation (and the services that depend on it) in the first place.

If we don’t take back the wheel soon, there may not be a future left to chart.

Paris includes a brief history of data centers but we should mention Moore’s (Gordon) Law, formulated in the early 1970s when he noticed that the bit capacity of DRAMS (dynamic random access memory) was doubling every 18-24 months. Despite constant predictions that this progress will have to end, the capacity of silicon chips continues to double making it the first exponential growth technology in human history. The Internet, data centers, AI would not exist without this exploding silicon technological growth.

The Internet as we know it today dates from 1995 when commercial restrictions were removed. An explosion of startup companies followed until the dot-com crash of 2000. The few survivors quickly amassed monopolistic dominance to join Microsoft whose monopoly was based on their control of the PC’s operating system. These included Google in search.

In 1997 Professor Ramnath Chellappa defined the  term cloud computing which was adopted as the marketing term for commercial data centers. Jeff Bezos of Amazon  built a large worldwide system of data centers (Amazon Web Services AWS 2002 and 2006) that could be marketed to other companies as well as support Amazon businesses. He was soon joined by Microsoft (Azure 2008) and Google (App Engine 2008).

In 2008, the sub-prime mortgage crash crippled much of the world economy, leading to a prolonged very slow recovery that central banks tried to stimulate through a monetary policy of zero interest rates. When the worldwide Covid-19 pandemic struck, the central banks continued their zero interest rate monetary policy. The 2008 global recession allowed these cloud data centers to be built with low cost financing and the operators were willing to build and operate these systems for years without profit.

Companies worldwide began to replace their in-house data systems with cloud computing and clouds became profitable for the first time. Over 29% of the global population uses personal cloud services, with higher regional rates such as 62% in the United Kingdom.

Netflix started its DVD rental business in 1998. In 2008 Netflix moved their system to AWS and by 2012 they were a fully streaming service on AWS. YouTube was acquired by Google in 2006. By 2008 Google was able to stream YouTube on their cloud.

In 1999, Nvidia released the GeForce 256, marketed as the world’s first graphics processing unit (GPU), which shifted geometric processing load away from the central processor (CPU). In 2006, Nvidia introduced CUDA (Compute Unified Device Architecture), allowing GPUs to handle general-purpose parallel computing tasks well beyond standard video rendering.  In 2012 Researchers used Nvidia GPUs to train AlexNet, a breakthrough neural network that sparked the modern artificial intelligence boom.

By 2020 OpenAI, Anthropic and others had adopted LLM (Large Language Models) and chatbots as their strategies and the age of hyper-scale data centers powered with massive numbers of GPUs began.

The book includes excellent chapters exposing the negative global societal impact of hyper-scale data centers on POWER, WATER, CLIMATE, AND CHIPS (the environment and societal costs of manufacturing and disposing of the necessary hardware for hyper-scale data centers.) The benefits to society of these data centers remains doubtful and unclear.

The ad (Wall Street Journal Apple ad Aug 1980) was indicative of an ideology forming around that time, particularly among those in the growing high-tech industry in Northern California, that brought together several strands of thought to create the tech-infused libertarianism that’s become synonymous with Silicon Valley. The Californian Ideology…combined technological determinism and a form of countercultural individualism with the neoliberal ideas of the Reagan era. In effect, the idea was that social progress would be found not through politics or collective action but through entrepreneurialism and the pursuit of new technologies, enabled by a government that got out of the way and let companies do what they wanted.

Reporters Harry Davies and Yuval Abraham (Aug 6, 2025 The Guardian) revealed that the commander of Unit 8200, Israel’s military intelligence agency, had met with Microsoft’s CEO Satya Nadella and received his support to use a segregated part of the company’s Azure cloud platform to store vast quantities of top-secret intelligence…internal records of the meeting showed Nadella supported the spy chief’s plans to move up to 70 percent of Unit 8200’s data onto Microsoft’s cloud.

Those revelations prompted Microsoft’s workers to occupy parts of the company’s headquarters in Redmond, Washington, and stage a sit-in at the office of President Brad Smith. Four workers were ultimately fired for participating in the demonstration.

The Edward Snowden revelations in June 2013 revealed the existence of an NSA program called PRISM. For a history of Silicon Valley’s relationship with the US government and pentagon see

The report (Greenwald and MacAskill June 7, 2013) detailed a secret program called PRISM through which the NSA gained direct access to data on the servers of those major tech companies. A leaked slide from an internal presentation about PRISM showed the agency had gained access to Microsoft servers in 2007. It added Google and Facebook to the mix in 2009, and Apple in 2012.

This is not a vision for society that was collectively or democratically decided on. Quite a small group of very wealthy and powerful people are crafting a model for the future that serves their interests and the bottom lines of their companies and that is increasingly coming at the expense of everyone else. It’s also one that people can be excluded from at a a moment’s notice, with catastrophic consequences.

Over time, that meant not just companies but governments and public institutions came to be effectively run on digital services and software controlled by US companies and hosted on servers within data centers that those companies owned. By 2025, US companies controlled 70 percent of the cloud market in Europe, compared with just 15 percent for European firms. Governments were deeply dependent on Microsoft’s Windows operating system and 365 services suite. Google held an important foothold of its own.

Even undersea cables, which had traditionally been the domain of telecommunications companies, were coming under the control of US tech companies. By 2023, Amazon, Microsoft, Google, and Meta had collectively invested in nearly fifty cable projects. At that time, 99 percent of internet traffic passed through the 486 cables, but not all cables were equal. Just two ears later, those four companies accounted for more than 70 percent of all used undersea cable capacity.

These dominant companies control infrastructure like data centers and undersea cables. They shape society through their influence and digital services they control to serve their own ends, with little regard for the wider ramifications. ..They’re spearheading the creation of a society of disconnection and dislocation where digital mediation is the rule and nonhuman relationships are normalized.

Through that latter half of 2025, there was growing consensus that the industry was spending far too much on data centers to justify the return they expect to receive. Bain & Company estimated that the annual revenue needed to pay for the buildout appeared to be $800 billion short, which called into question how much the company’s stocks had soared. Torsten Slok of research firm Apollo found the top ten companies in the S&P 500 were more overvalued than during the dot-com bubble in the late 1990s, while MacroStrategy Partnership estimated the AI bubble was seventeen times larger than the dot-com bubble.

What really pushed (of The Big Short fame Michael) Burry to bet against it (AI companies) was seeing how the companies seemed to be actively creating the conditions not just to overvalue every piece of the AI value chain, but also to increase the risk of a collapse.

These circular deals became a feature of the generative AI market, making it look like money was flowing into companies engaged in making the technology a mass-market product, when it was really just flowing between the same small groups of important players. Both Google and Amazon invested billions in Anthropic, ensuring that money would flow back to them in form of contracts for their cloud divisions. Microsoft’s investment in OpenAI was even more direct on that front: Much of its $10 billion investment was in the form of cloud credits instead of cash. When OpenAI agreed to buy an eye-watering number of GPUs from Nvidia’s chief competitor, AMD, the deal included the option for OpenAI to buy 10 percent of the company (AMD).

Burry believes GPUs are operating on a two-to-three year product cycle, and that the major players in the AI race will underestimate the depreciation of their assets by $176 billion between 2026 and 2028.

…McKinsey estimates cumulative spending on data centers could get as high as $6.7 trillion between 2025 and 2030. So where is all the revenue coming from to justify those expenditures?

At around three in the morning on the US East Coast (Oct 20, 2025), the largest cloud operator in the world ran into trouble, kicking off a widespread outage that would last more than fifteen hours. Amazon later explained that two automated programs had tried to overwrite the same DNS entry at the same time. Instead of leaving behind one of the versions, the entry was simply left blank.

DNS or Domain Name System, is essentially like the phone book of the internet, so once the record was empty, the computer’s servers couldn’t figure out where to find the connections they needed. They couldn’t make those essential calls out to other parts of the system. That flaw proceeded to bring down a series of other services too, particularly DynamoDB, where many customers store their data. The extent of the outage is part of the reason it took so long for everything to come back online. It wasn’t just one system that was affect; that one flaw had taken out many others with it.

After the dust had settled and everything was back online, that one mistake in the bowels of the AWS network was estimated of have cost upward of half a billion dollars. Even more than that, it revealed a deep vulnerability in an infrastructure that has become so central not just to the internet but to a modern, connected society, compounded by outages in other major services in the months that followed.

While researching this book, I performed a number of google searches in AI mode. This link gives the Gemini formatted responses with reference links removed.

Mission Economy: A Moonshot Guide to Changing Capitalism, Mariana Mazzucato, 2021

As this book goes to press (during COVID-19), the problems being faced by both the USA and UK are the result of forty years of weakening ability to govern and manage — fueled by the ideology that government needs to take a back seat and only come in to fix problems when they arise. (like bail out the financial sector in 2008).  A public management creed that belittles the ability of government to act effectively and promotes privatization has fostered much outsourcing of government capacity to the private sector and a relentless but misguided focus on static measures of efficiency, leaving governments with fewer options and even latching onto unrealistic technology panacea such as artificial intelligence or ‘smart cities’. It has also led to lower investment in public capabilities, the loss of institutional memory and an increased dependence on consulting companies, which have benefited from billions in government contracts.

In the UK, the government outsourced health contracts worth £9.2 billion in 2018 alone. Over 84 percent of care home beds are in privately owned homes, and 50,000 of those are beds in homes run by private equity (PE) companies those ultimate aim is profit, not care…The total value of the public-health grant in the UK — which enables local authorities to provide vital health care and preventative services — has been declining in real terms, from £4 billion in 2015 to £3.2 billion in 2020/2021, a decrease close to £900 million.

I call this (Apollo) different way of doing things a mission-oriented approach. It means choosing directions for the economy and then putting the problems that need solving to get there at the centre of how we design our economic system. It means designing policies that catalyze investment, innovation and collaboration across a wide variety of actors in the economy, engaging both business and citizens. It means asking what kind of markets we want, rather than what problem in the market needs to be fixed. It means using instruments such as loans, grants and procurement to drive the most innovative solutions to tackle specific problems, whether those be getting plastic out of the ocean or narrowing the digital divide. The wrong question is: how much money is there and what can we do with it? The right question is: what needs doing and how can we structure budgets to meet those goals.

Changing capitalism means changing both how government is structured and how business is run — and how public and private organizations, and relationships between organizations, through a notion of ‘purpose’ is the key to a mission-oriented approach.

By 2018, private debt to GDP reached 150 percent in the USA, 179 percent in the UK, 200 percent in France and 207 percent in China — all substantially higher than levels at the turn of the century.

And much of business has been plagued by a dangerous combination of low investment, short-term management and high rewards to shareholders and company bosses.

Today, China, the leading authoritarian economy, remains weighed down by inefficient and heavily indebted state industries, a banking system with huge ‘zombie’ loans, an ageing population, and the massive task of shifting the economy away from excessive export dependence and towards greater domestic consumption…But a central planning model is not likely to be one that will be able to take on the bold reforms to public and private collaboration that this book envisages.

To grasp the true scale of this challenge, it is important to understand that the issues described above are the consequences of deeper forces that together have led to a dysfunctional form of capitalism. There are (at least) four key sources of the problem: (1) the short-termism of the financial sector, (2) the financialization of business, (3) the climate emergency, and (4) slow or absent governments. In each, the way that organizations are structured and how they relate to each other are part of the problem. Their restructuring must, therefore, be part of the solution.

Over the same period (2008-2019), six of America’s biggest airlines spent an average of 96 percent of their free cash flow on stock buybacks — the aircraft manufacturer Boeing spent 74 percent of its free cash slow on stock buybacks — which didn’t deter these companies from asking for federal government help when the COVID-19 crisis struck.

The 2019 Intergovernmental Panel on Climate Change report argues that we have just ten years left until climate breakdown is irreversible. COVID-19 made us even more aware of the fragility of our environment and an economic system dependent on it. The climate emergency is very much a result of the four other problems in finance and business: fossil fuels still dominate our energy sources, industries are too carbon intensive, the financial sector has fed the fossil-fuel-driven economy, and government is nurturing this dysfunctionality – in 2019, extraordinarily, subsidies to fossil-fuel companies were estimated at $20 billion a year the the USA and an enormous €55 billion a year in the European Union (EU). Of the COVID-19 recovery funding allocated to energy companies by G20 governments, 56 percent has been handed to fossil-fuel projects, equivalent to $151bn (£119bn).

The problem is not ‘big government’ or ‘small government’. The problem is the type of government: what it does and how.

Mission Economy is about how government must change from within in order to deliver on ambitious outcomes, as well as how it must change its interaction with other actors. The current state of the typical instruments used by governments, such as taxation, fiscal policy and monetary policy, is rudderless. There is no systemic directionality towards de-financialization or sustainability. No attempt to change the relative cost structure through the tax system and other instruments to gear investment towards elimination of waste, reducing materials and energy use and pollution. Why are we not seriously taxing the environmental ‘bads’ and favouring ‘green’? Why are capital gains taxed less than income from work? Why is there no clear benefit for investing long-term rather than short-term? Why is nothing done to stop doubtfully legal high-frequency trading as well as the harmful practice of excessive share buybacks? Why is there no attempt to redesign the welfare state to respond to the new conditions created by the digital technologies?

But once a proper direction is systemically applied to the whole range of policies, leading to synergies in innovation and investment behaviour across the economy, government still needs to do much more. It must transform itself into an innovating organization with the capacity and capability to energize and catalyze the economy to be more purpose-driven.

The Common Good Economy; How to Make Capitalism Work for Us All, Mariana Mazzucato, 2026

Practical men, who believe themselves to be quite exempt from any intellectual influences, are usually the slaves of some defunct economist. Madmen in authority, who hear voices in the air, are distilling their frenzy from some academic scribbler of a few years back. – John Maynard Keynes (1883-1946) The General Theory of Employment, Interest and Money (1936-1973) Also quoted by Susan Neiman in Call It Evil; Understanding the Trump Era 2026.

The book gives a brief summary of key enlightenment figures Thomas Hobbs, John Locke, Jeremy Bentham, David Hume, and Adam Smith, Hegel, Marx, Engels, and John Stewart Mill:

By the late nineteenth and early twentieth centuries, the emphasis on individual freedoms had consolidated into a dominant liberal framework, placing individual autonomy and self-interest at the center of political thought… Mill (1859) championed individual autonomy, resisting the authority of moral or religious communities that sought to impose goals on others. Similarly, (John) Rawls (1971-1993) maintained that the state should remain neutral among competing conceptions of the good life, focusing instead on securing the freedoms and opportunities necessary for individuals to pursue their own well being without coercion…Rawls derived two principles of justice. The first guarantees equal basic liberties for all citizens including freedom of speech, conscience, and the rule of law. The second, known as the difference principle, permits social and economic inequalities only if they benefit the least advantaged and arise within a context of equality and opportunity.

In a section entitled Lessons on collective intelligence from biology Mazzucato highlights:

Parasitic ecosystems benefit one organism at the expense of others, whereas symbiotic or mutualistic ecosystems allow all organisms to benefit while keeping each other in check. When balance in the ecosystem is disrupted by one organism becoming disproportionately powerful, the ecosystem is threatened with collapse.

Similarly forest ecologist Suzanne Simard (2022) has studied how trees communicate via underground mycorrhizal fungal networks, which she has called the ‘woods wide web’. She challenged previous work on trees that view forests as competitive environments and argues instead that trees often collaborate to support each other’s health and growth.

Scholars of the ‘collaborative economy’ study the conditions that allow communities to flourish, emphasizing community building, inclusivity, and sustainability…Relatedly, work on ‘collective intelligence’ emerging from fields such as sociology and psychology , captures the idea that people working together can solve problems better than they can individually, resulting in continuous experimentation and innovation…Moreover, this intelligence is even greater when there is social diversity, as a growing body of research has shown that diverse groups make better decisions.

In philosophy, this narrow (Eurocentric) view obscures the fact that profound philosophical traditions around the common good have long existed outside the West, including Confucianism in East Asia, Islamic philosophy across the Middle East, Ubuntu in sub-Saharan Africa, Buddhist and Hindu thought in South Asia, and the diverse worldviews of Indigenous peoples across the Americas, Oceania and beyond.

Linux was created in 1991 by Linus Torvalds while he was a computer science student at the University of Helsinki in Finland because Bell Lab’s originated Unix based computers were too expensive. Linux is open source and became the default operating system chosen as Internet servers were globally deployed. Today’s Linux’s descendant is Ubuntu, also open source and attracting large numbers of open source applications. This blog was written on an Ubuntu laptop.

Many Indigenous worldviews understand the self as emerging from relationships with community, the land and the spiritual world, rather than as something defined in isolation…By rejecting this separation, Indigenous communities view the individual and the collective as mutually sustaining, reflecting a more holistic understanding of the interdependence of the social, ecological, and spiritual dimensions that sustain life.

For example, in Ecuador, the Quechua concept of sumak kawsay – translated into Spanish as buen vivir ‘good living’ – reflects a worldview (cosmovision) in which the good life is not achieved through individual accumulation or consumption, but through harmony with nature and collective coexistence. Ecuador formally updated its Constitution in 2008 to incorporate buen vivir as a foundational principle of national development and governance.

A major advance in social economics came with Amartya Sen’s capability approach (1979). This framework shifts the focus from utility-based welfare to what individuals are actually able to do and be — what Sen calls ‘functionings’. Functionings encompass basic conditions such as nutrition and health as well as more complex aspects like self-respect and meaningful participation in community life. An individual’s ‘capability set’ — the range of functionings they can achieve — reflects their freedom to pursue a life they value, shaped by both personal characteristics and broader social, economic and institutional conditions.

Communitarian, biological and Indigenous perspectives reveal that individuals do not exist in isolation, but within interdependent social and ecological networks. Identity, wellbeing, and even generation of value in the economy are inseparable from these relationships. This challenged the idea of value as something produced only by individuals and instead points to collective value creation, in which markets are shaped by shared institutions and common objectives.

The distinction biologists draw between symbiotic and parasitic ecosystems provides a useful analogy for rethinking economics. A symbiotic economy requires attention to the quality of relationships which are shaped and governed through a wide range of mechanisms such as property rights, water rights, intellectual property, and data flows. These underlying arrangements determine whether economic systems foster cooperation or enable exploitation.

Much of mainstream economic thought has ignored such interdependence, treating humans as isolated agents and markets as neutral mechanisms. This perspective leaves little room for the social and ecological dimensions of value creation and confines collective action to addressing failures after they happen. By focusing on the way relationships are structured from the start, both within communities and with nature, we open the possibility of shaping our economy to actively advance the common good.

By contrast, former US Secretary of Labor Robert Reich (2018) in his book The Common Good, explores the idea of the common good in relation to what he believes to be the erosion of shared values and mutual obligations in American society. He argues that the historic US commitment to the common good — stemming from the Declaration of Independence and the US Constitution, and expressed through ideals such as the rule of law, democratic institutions and a commitment to truth — began to erode in the late 1970s. He perceives three major structural breakdowns have driven this shift; the degradation of political into partisan competition, the retreat of corporate responsibility in favor of shareholder primacy, and distortions within the market that encourage exploitation over fairness. Together, these dynamics have weakened public trust, accelerated economic inequality and shifted attention away from collective wellbeing towards private gain.

To restore the common good, Reich calls for a renewal of civic life grounded in trusteeship, where leaders see themselves as stewards of the common good. He emphasizes reclaiming the social use of honour and shame to condemn corruption and celebrate virtue, rebuilding a culture of truth telling in public life, and reviving civic education to instill responsibility and democratic values in future generations. In doing so, he envisions a society in which the common good once again guides the collective. Yet, like the communitarians, Reich does not provide a framework for ensuring that the common good actively shapes how the economy operates at its very core: the role of business, the role of government policy beyond fixing markets, and how the two interrelate.

A common good economy requires embedding economic objectives within the fabric of social relationships and rethinking how all actors — individuals, public institutions and private firms — relate to one another. This demands removing it from a notion of market failures; it must instead be about how to shape markets and the economy, putting the common good at the centre of all economic relationships. And it also cannot be about non-market activities. It must go to the centre of how economics and markets work. A common good framework for the economy is about aligning goals, incentivizing collaboration, fostering collective intelligence, and ensuring that all participants share knowledge, risks and rewards. It moves beyond treating markets as neutral arenas for individual maximization, aiming instead to design systems that actively cultivate cooperation and shared wellbeing to make the common good possible.

To create a new economics of the common good, we must thus shift away from a focus on maximization exercises towards collective value creation, where interactions between business, governments and other organizations are viewed not as mere transactions but as relationships to be guided through principle of reciprocity.

By understanding the role the state played in forcing markets into existence, we understand that there are no deterministic market laws, but rather markets are outcomes of governance decisions. Similarly, Evans (1995) concepts of ’embedded autonomy’ show how states actively organize and stabilize markets through structured engagement with firms while retaining independence.

Market shaping can lead to more inequality, if the state is working for the interests of capital, as Marx argued. Or it can lead to more pollution, depending on the governance of companies, as well as the tax system. The question then becomes how to have the kind of government policy and governance structure that shape markets to create a more inclusive and sustainable capitalism. Outcomes orientation in this view means that instead of just viewing the bad things that arise in the economy as ‘externalities’ we can better understand how those bads are produced by the way markets work even in ‘perfect’ form, indeed through the very logic of the market.

In her book The Code of Capital, Katharina Pistor (2019) talks about how inequality is coded in the law. Without un-coding, inequality remains. Marx insisted that capital was a social relationship, based on capital exploiting labour. Pistor argues that capital isn’t a natural or physical thing, but a legal construct crafted through what she calls ‘legal coding’; the selective deployment of legal tools (property rights, contracts, trusts, and bankruptcy law) that give an asset priority, durability, universality and convertibility, enabling it to generate wealth and be enforced globally. Private lawyers act as the architects of this code, deliberately shaping financial instruments, corporate structures and even intangible assets to concentrate wealth among the legally sophisticate, perpetuating inequality across countries and generations.

Unless these legal frameworks are unraveled, it is nearly impossible to change the rules of the game that produce inequality and disempowerment. Attention to legal relationships must be matched by attention to all relationships at the interface between public and private sectors, from property rights to procurement contracts to bailout schemes. Only by redesigning these structures can we begin to foster the mutualistic ecosystems necessary for an economics of the common good.

If businesses are wed solely to maximizing shareholder value, then we get specific market outcomes: a fictionalized economy where what matters most are quarterly returns and stock prices…But if we build corporate models that genuinely align long term success and value creation with social and environmental value, businesses can become powerful engines for good growth.

“Global economy watchers have inundated the internet with historical parallels for the triple financial bubble at hand, inflated by hopes and dreams for artificial intelligence and cryptocurrencies, and previously unimaginable levels of borrowing.” World Economic Forum Nov 2025

Over the past forty years or so the stock market has become increasingly decoupled from the real economy. The growing size of the financial sector and that financialization of our economics point to the need to reduce finance’s share of the economy, while making sure that the finance that exists truly serves it.

Furthermore, finance is often very short term, making trillions from millisecond trades that are just about buying and selling existing assets instead of creating new ones.

Of all the vaccines produced during the (Covid 19) pandemic, only one embodied an outcomes-oriented approach in the design of the public-private partnership; the collaboration between Oxford University and AstraZeneca (sometimes referred to as ‘Vaxzevria’). In this model, early stage developments by Oxford were transferred to AstraZeneca for final development, manufacturing and distribution. Within this arrangement, the UK government embedded conditionalities in the Vaxzevria Supply agreement, requiring vaccines to be delivered at cost. As stated in the Agreement: ‘AstraZeneca shall use Best Reasonable Efforts to mitigate and reduce the Cost of Goods during the Term of this Supply Agreement. These terms significantly enhanced the UK government’s ability to guarantee vaccine access for all UK citizens.

After the pandemic, different attempts were made to learn the lessons on how make sure Global South countries are able to produce their own vaccines: for example, the creation of the mRNA Vaccine Technology Transfer Program. Its main purpose is knowledge sharing to increase the capacity of the Global South countries to produce their own vaccines. By distributing technology across multiple production sites in the Global South, the programme seeks to decentralize and diversify mRNA vaccine manufacturing, counter extractive intellectual property practices, and ensure that early-stage, high-risk public funding is conditional on knowledge sharing and equitable access.

Another example of knowledge sharing is the International Treaty on Plant Genetic Resources for Food and Agriculture often referred to as the Seed Treaty. Established by the UN Food and Agriculture Organization and adopted in 2001, this global agreement created a multilateral system for the sharing of plant genetic materials crucial to food security and sustainable agriculture. The Seed Treaty enables countries to freely access a pool of over sixty important food crops and forage plants, including staples like rice, wheat and maize, which have been cultivated, bred and improved upon over centuries by farmers across the globe.

From a common good perspective, the Treaty represents a shift in governance from a system based on ownership and enclosure to one based on stewardship and reciprocity. It reflects the co-creation and participation we discussed in the previous chapter, ensuring that smallholder farmers and Indigenous communities – who are often the original custodians of these genetic resources – are included in decision-making and benefit-sharing processes. Moreover, it promotes transparency and accountability, the final element of the common good compass, in how plant knowledge-sharing agreements, when designed around public purpose, can reshape economic incentives and legal norms toward more inclusive, sustainable outcomes. It also reinforces the point that innovation ecosystems should not only produce technological progress but must also institutionalize fairness in access and distribution – something that the Seed Treaty operationalizes through international law.

Digital Public Infrastructure (DPI) – a layer of infrastructure that sits above hardware and internet protocols and underpins the functioning of modern markets and societies – is emerging as the defining features of this century, like roads, electricity and water systems. DPI forms the backbone of our collective life. For much of the past two decades, the digital agenda had focused narrowly on service delivery, administrative efficiency and enabling private sector innovation. But today, DPI is increasingly recognized as the ‘railways of a digital economy’ – a shared infrastructure that connects people, supports the delivery of essential services and rights – and promotes inclusion.

What is needed is a clearer understanding of what constitutes public value in the context of DPI and how that value is intentionally created and sustained. Without this framing, it is impossible to ensure the transparency and accountability of the most important technological changes of the twenty-first century.

Two lenses are commonly used to assess DPI’s public value creation: attributes and functions. Attributes refer to the technical features that make infrastructure scalable, flexible and open – such as interoperability, modular design and open source code. These features support dynamic efficiency and reduce fragmentation across digital ecosystems. Functions, on the other hand, refer to the roles DPI plays in society; fostering community and social relationships, enabling economic activity, guaranteeing essential capabilities and enhancing overall quality of life.

While both are important, they are not sufficient. A system can be technically open and functionally useful yet still fall short of advancing public goals or being governed in ways that promote the common good. Instead, a third lens is needed: the common good elements discussed in this book, which focus not only on outcomes but also on the processes and institutional arrangements through which value is shaped, distributed, and sustained.

Collective learning and knowledge sharing imply that DPI should be conceived of as a site of continual experimentation and feedback. Open source platforms such as the Modular Open Source Identity Platform (MOSIP) enable global collaboration and localize adaptation. By supporting shared infrastructure that evolves through distributed contributions, such models institutionalize collective intelligence and align innovation with long-term public value.

 

For Instance, the Bezos Earth Fund, established by Jeff Bezos has made important investments in climate adaptation. Yet the fund is funded by wealth generated through deeply problematic business practices. Amazon, the company Bezos owns, has relied on tax avoidance and labour exploitation to accumulate massive profits, while also producing 68.25 million metric tonnes of CO2 in 2024 – comparable to a mid-sized European country – and its carbon footprint is expected to grow further due to its AI data centres. While philanthropic initiatives like the Bezos Earth Fund provide some support for addressing climate change, they cannot offset the harm caused by the company’s core business practices.

The World Bank’s Mission 300, which aims to provide electricity to 300 million people in Africa by 2030, combines public and private investment to expand power generation and transmission, scale up decentralized renewable solutions such as mini grids, and improve affordability and regional integration. It also emphasizes community empowerment, particularly for women, by enable access to clean energy and productive uses that foster local enterprise.

In many ways, Mission 300 aligns with several principles of the common good compass. It promotes co-creation through collaboration with governments and local partners, supports collective learning via open data and shared platforms, and seeks fairer reward sharing by linking investment to social outcomes rather than short-term returns. It also emphasizes accountability and transparency through public reporting and feedback mechanisms, indicating a move towards more participatory governance.

These considerations are important for thinking about how to govern artificial intelligence (AI) around common good principles. A technological breakthrough of epochal proportions, AI could and should be used for the common good. The uncomfortable truth, however, is that it is quickly going in the wrong direction. It depends on access to the gatekeeper’s cloud computing platforms. Three companies – Amazon Web Services, Microsoft Azure, and Google Cloud – control around 70 per cent of the cloud’s market share. This oligopolistic ownership structure has turned AI into a venous engine of rent extraction and a drain on knowledge from the public sector, as the private companies use their grossly excessive trillionairs incomes to lure top talent away from public labs, universities and the public service with very high salaries. Meanwhile, data privacy is being breached, misinformation is rampant, and the divide between unskilled and skilled labour is widening. AI platforms also consume enormous amounts of energy, water and minerals, exacerbating climate change, the water crisis and other environmental challenges.

Yet, like previous technological revolutions, today’s AI breakthroughs, from voice recognition to large language models were built on the shoulders of public investments. AI models are largely trained on the creative work of writers, singers, artists, journalists, illustrations and more whose creations are used without permission, and who almost always receive no compensation. Simply tightening copyright protections, as many critics have proposed, will in and of itself be insufficient to correct this dynamic. We should instead treat collective knowledge in the age of generative AI as a common good and collectively fund (and benefit from) its production. Like clean water or vaccines, the creative commons should be accessible to everyone, with risks and rewards equitably distributed among all actors.

AI is still in its infancy, though developing fast. Nobody can truthfully say what direction it will take or what its impact on society will be. We can say, however, that the best approach to AI governance would be to design a public-oriented structure that embodies the five elements of the common good compass to steer AI innovation towards public value creation..

First, AI development should be guided by a clear public mission to advance collective wellbeing, social justice and environmental sustainability, rather than serving the short-term profit motives of a few dominant corporations. This mission must be pursued through co-creation and participation involving public institutions, universities, civil society, citizens and workers across sectors. Such collaboration would ensure that the design deployment and regulation of AI reflect diverse perspectives, public needs and democratic accountability, rather than being shaped exclusively by private commercial interests.

Knowledge sharing and open access to data, models and research should replace the current reliance on secrecy and excessive patenting. AI advances funded by public resources should remain open and interoperable, allowing for collective learning and the diffusion of innovation across the public and private sectors. Moreover, risk and reward sharing must be embedded into the AI ecosystem to ensure that value generated from collective inputs benefits society as a whole. AI should be publicly supported with clear conditionalities so that outputs remain in the public domain and serve public purposes.

Finally, transparency and accountability must underpin all aspects of AI governance. Independent oversight bodies, open metrics and publicly accessible evaluation frameworks would make it possible to track AI’s social, environmental and ethical impacts, challenge failures, and hold both public and private actors to account.

The Triple Bubble Threat to the Global Financial System