In Johannesburg, 3,000 African business and government leaders gathered for Google’s first-ever Africa Cloud Summit, opened by South Africa’s President himself. In orbit, serious analysts and investors were discussing whether the next data centre might not be on Earth at all. In Singapore, a small island nation quietly refined the planning document that ensures every pipe, cable, and power line will be in the right place fifteen years from now.
The thread connecting all of it is the same question this newsletter has been asking for months, in different ways, in different languages. Who gets to build the infrastructure of the AI era, and who gets to benefit from it? This week, the answers were more interesting than usual.
Google Just Made Africa’s AI Infrastructure Future a Little Less Theoretical
On July 1, 2026, Google hosted its inaugural Africa Cloud Summit at Johannesburg’s Sandton Convention Centre, opening with remarks from South African President Cyril Ramaphosa and attended by 3,000 business leaders, developers, and policymakers from across the continent. Google announced five major initiatives covering new infrastructure, applied AI research, startup funding, and digital skills, anchored by its Johannesburg Cloud Region, which is projected to contribute $90.6 billion in additional economic output and support 314,900 jobs by 2030. Among the most significant announcements was the launch of Africa’s first applied AI lab, based at the Accra AI Community Centre in Ghana, pairing African founders with Google researchers and providing access to Google’s latest models to build solutions for uniquely African challenges.
Google has already surpassed its earlier $1 billion commitment to Africa’s digital transformation ahead of schedule, and the summit’s theme, “Building for Africa with Google Cloud,” reflected a deliberate shift from experimentation to deployment. African enterprises, according to Google Cloud VP Maureen Costello, have moved past the AI experimentation phase and into the agentic AI reality of 2026, where autonomous systems are delivering measurable business value in real environments. The moment a continent of 1.4 billion people stops being a pilot market and starts being treated as a deployment priority, the global AI infrastructure story changes in ways that cannot be undone.
1. The AI Infrastructure Arms Race Is Now Looking at Outer Space
The Italian Institute for International Political Studies and Forbes both published analyses this week examining serious investment interest in space-based data centres, with SpaceX’s post-IPO capital position making orbital compute infrastructure a commercially fundable thesis for the first time, as Earth’s power, water, and land constraints increasingly bite into terrestrial AI deployment timelines.
If even a fraction of the capital currently flowing into terrestrial AI factories shifts toward orbital infrastructure, it will create an entirely new asset class in space, one with no precedent for how it gets regulated, taxed, or governed, and those framework decisions will be made in the next three to five years before most governments realise they need to have an opinion.
The country that writes the rules for space-based AI infrastructure will hold a regulatory advantage that lasts longer than any single data centre on the ground.
Read the ISPI analysis on the future of AI infrastructure from orbit to ocean.
2. Singapore Shows the World What Infrastructure Planning Actually Looks Like
Singapore’s Urban Redevelopment Authority published new detail on its National Infrastructure Plan this week, a 15-year rolling framework that brings together every government agency and utility licensee annually to align underground and above-ground infrastructure plans before a single project breaks ground, with a Major Infrastructure Workflow introduced in 2024 extending this coordination from planning through implementation across key development corridors.
In a world of expensive infrastructure blowouts, politically driven announcements, and 35-percent cost overruns on projects that skip independent review, Singapore’s NIP is the clearest working proof that long-horizon, cross-agency, upstream coordination is not a bureaucratic luxury but the single most cost-effective infrastructure policy a government can have.
Every infrastructure minister who complains about cost overruns should spend a week in Singapore learning how to prevent them.
3. Energy’s New 2 Percent Is Reshaping the Entire Grid
The Hill published analysis this week tracing how data centres, once a negligible fraction of total electricity demand, have grown from roughly 2 percent of US power consumption to become the dominant driver of new grid investment decisions across both Europe and the United States, with transmission infrastructure, power generation, and industrial development now being planned around data centre load curves rather than the other way around.
This inversion matters because grid infrastructure built to serve industrial and residential demand over thirty-year planning horizons is now being retrofitted to serve a technology sector whose power requirements double every two to three years, creating a structural mismatch that is showing up as interconnection queues, grid congestion, and rising electricity prices for consumers who never signed up to subsidise an AI buildout.
When a 2-percent load becomes the primary driver of how a nation’s entire energy infrastructure gets planned, something has shifted in the power relationship between technology and society that deserves a lot more public conversation than it is getting.
This post is public so feel free to share it.
1. The WEF Argues Space Access Must Stop Being an American Monopoly
The World Economic Forum published a compelling analysis this week arguing that space infrastructure, already quietly essential to farming, navigation, communications, and climate monitoring for billions of people, must become genuinely multi-polar, as the concentration of launch capability and satellite infrastructure in a handful of US companies creates fragility that the world cannot afford when those systems fail or are disrupted. This matters for infrastructure planners because the satellite layer is now as foundational to daily life as the power grid, and it has none of the regulatory diversity that energy infrastructure has accumulated over a century.
Read the WEF analysis on why the future of space access must be multi-polar.
2. Melbourne Is Running a Live Experiment in AI-Managed Energy Infrastructure
IEEE Spectrum reported this week on Melbourne’s deployment of AI-driven energy management systems to handle the compound load demands of electrification, AI compute growth, and extreme heat events on a grid that was not designed for any of them simultaneously, with the experiment being described not as a future pilot but as a present, system-level requirement for any city grid facing AI-era demand. Watch this closely, because Melbourne’s results will be referenced by every grid operator in every hot city trying to understand what coordinated AI energy management looks like at urban scale.
3. Forbes Says the Second Space Race Will Be Won by Infrastructure, Not Rockets
Forbes published a pointed analysis this week arguing that the decisive factor in the emerging competition for space-based resources and positioning will not be launch capability or scientific ambition but the ability to build robust, lasting infrastructure that supports human presence and economic activity beyond Earth’s surface, citing the history of every previous territorial expansion as evidence that explorers get naming rights and infrastructure builders get the economy. For investors and governments thinking about the next fifty years of capital allocation, the argument is worth sitting with.
$90.6 Billion
That is the projected economic output that Google’s Johannesburg Cloud Region alone is estimated to contribute to South Africa by 2030, a figure that makes the case, more concretely than any policy paper could, that digital infrastructure investment in Africa is not development charity but one of the highest-return bets available in the global economy right now.
Watch for the first results from Africa’s applied AI lab at the Accra AI Community Centre, with applications closing August 31, 2026, because the quality and ambition of the founders who emerge from that first cohort will tell us whether Google’s Africa AI infrastructure bet is building a genuine innovation ecosystem or a well-funded outpost of Silicon Valley thinking applied to African problems from the outside.
No posts
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.