The map places the ten markets. Strata reads across them, seven cross-cutting layers that turn ten markets into one thesis.
The cross-cutting lens. Four infrastructure archetypes, and the non-negotiable conditions each one demands.
Every market, ranked. How the ten hotspots score against each other, and against the conditions on the ground.
The industry, surveyed. Conviction and caution from the people building, financing and powering the sites.
Seven live instruments. Where demand, power and reality are weighed against one another, read across every region.
Where the accelerators actually land. A GPU trade-flow read: bill-to against ship-to, and the gap between announced power and the chips to fill it.
Who funds the build, and how the capital stack is structured. The cost has inverted toward silicon, and private credit is carrying the largest external share.
Where the next hotspot lands. The thesis the whole report builds toward, and what to watch from here.
Beyond the markets and the cross-cutting layers, original analysis – longer reads and outside voices that pressure-test the thesis.
Andy Burnham reaches Number 10 having spent a decade making digital infrastructure a question of where it lands, and for whom.
Winning the argument, losing the room: how clip farms made data centres the internet’s favourite villain, and why being right stopped being enough.
The argument about data centres is being fought with figures that do not survive contact with the engineering. Four load-bearing claims, checked against the record.
From the AI PC on the desk to the hyperscale floor: how enterprise compute is stratifying, and why the value moves to the seams.
How the data centre’s hunger for memory is repricing every device you own – the cascade from a server hall to your checkout.
Power is the constraint everyone names; people are the one that quietly decides whether the power is ever switched on.
A robot is a walking edge device, and a fleet is a distributed data-centre problem. Six machines through the digital infrastructure of 2036, and what happens when it runs short.
The other half of the build. Which AI and digital-asset models become fundable, compliant workloads – read with Kerrigan and Stanford of CMS.
Complete is a construction status. Proven is an assurance status. The seams where risk accumulates, and the commissioning crunch coming for the build.
The outside voices behind the read – operators, lawyers, advisors and engineers, named and on the record. Their framing is theirs; the data and the read are the desk’s.
Read the constraint the country actually has, then pay for it yourself.
The operator who built TelecityGroup, on why power replaced demand as the constraint and what the customer does to your cost of capital.
Yes it can be built, but who buys it? Thirty years of delivery on offtake, the cost spread between hyperscalers, and what the industry should learn from bitcoin miners.
Why a 2.2 GW platform holds the connectivity hubs and builds the new capacity somewhere else entirely.
Complete is a construction status, proven is an assurance status. On the seams where risk accumulates between them.
Scores the delivery gap below the median, then argues the measurement is wrong. Why the capital went to build-to-suit, and why inference makes colocation matter again.
The licences bind in the centre. Fifteen hundred kilometres north, the wire binds instead.
The lawyer’s read on Europe’s build, where the rulebook now sits on the term sheet.
Scores the Singapore spillover thesis at 33 of 100, and argues Indonesia lost the AI cycle on policy rather than power.
The adviser’s read on who underwrites the build, and why neo-cloud credit quality is the discipline underneath the demand.
America has announced 780 GW of data centre against a 759 GW national peak. Two seats: the terrestrial gap in the United States chapter, and where the argument goes when the ground runs out of connections.
An industrial end user, reading the build from the customer side. The production process defines the need; technology expands how it can be met. Start at the other end and the process gets adapted to the solution.
PUE has reached its physical limits and the best sites run at ten to fifteen per cent overhead. The room that remains is inside the rack, not the shell.
The pipeline is filling and the bench is not. Entry level is improving; the bottleneck is a leadership layer that takes fifteen to twenty years to build and is not being measured.
The second water opinion, read from the pump side. Water is acute in some regions and secondary to power in most, and the fix is where you source, reuse and treat it.
Water is local, seasonal and political where power is a question of capacity. What a data-centre water footprint actually contains, and the seven ways published numbers mislead.
North Africa’s IaaS pioneer, on why the grid, not the chips, is the binding constraint.
Upstream of the build, where AI and digital-asset models either become fundable, compliant workloads or die in the lab.
The compute is built; the organisation is not. The demand may not arrive in the form the build assumes, and the constraint that decides it is institutional rather than technical.
How much candidate demand survives the regulation and the failure rate. Thirty-year assets are being built for tenants who may not be licensed or solvent in five.
Redundant urban buildings and stranded power instead of new campuses, with resilience in the network rather than the building. Scores compute leaving the planet at zero.
The capital read from inside the documents. What lenders demand before money moves, from power diligence and borrowing-base tests to what a GPU is assumed to be worth in year four.
A practitioner’s read on what actually decides a Gulf data-centre deal, against the grain of the questions.
Why demand and sovereignty, not announcements, decide whether Africa hosts its own compute. Africa AI for the good of all.
Africa is not ten years behind Europe but on a different trajectory, compressing three waves into one decade. Nine files on design, capital, connectivity and talent.
The only public-sector seat in the register. Where the state should lead, where it should stop, and why the south has the climate but not the grid.
Zero out of a hundred toward permitting. The constraint is power, the moat is land, and the German renewable obligation is almost impossible to meet.
Not a generation problem, a transmission one. Africa skips the density race, so water never becomes the constraint. Twelve questions, across two chapters.
Brazil is not short of power or capital. It is short of predictability. Ten files on the tax stack, the connection queue and why capital underwrites precedent.
The blocker is legal certainty, not power. And the announcement of ReDATA may have done more damage to the market than its delay.
Brazil is not short of generation. It is short of the step between authorised and connected, and a good share of the queue is paper.
Cost, not schedule, is what threatens a Latin American build. Why a regional benchmark rarely works, and where the overruns actually come from.
Why Gulf schedules slip at handover rather than on site, and why the binding constraint is commissioning talent rather than megawatts.
Standby has stopped being backup. The supplier read on on-site generation, bridge power at scale, and why the interesting move is financial rather than technical.
Land, power, cooling, connectivity. The fourth is listed last and is what breaks a site after it is built.
The leadership gap is a supply problem, not an attractiveness one. Which chair takes longest to fill, who translates from outside the sector, and what boards get wrong.
A security-first discipline for the AI build, from a global cybersecurity voice.
Whether announced UK capacity can actually be delivered, read against the grid and the supply chain.
Whether a UK build earns its local consent, now that planning has moved from local veto to national scrutiny.
A power-first read on what it takes to actually energise a Gulf campus, contributed in a personal capacity.
The capital view on Africa’s build: what gets underwritten, and what waits.
The connectivity lens on Brazil: why demand is outrunning the fibre, towers and power to support it.
The view from the allocators: which pension funds and insurers are mandating digital infrastructure as an asset class, and how fast the institutional money is actually moving.
The view from the regional front line on cooling, density and the Brazilian build.
A decade of Chile’s build, read from the front line of the market.
Every breakthrough begins with silicon: reading the chip cycle that sets the pace, and the price, of the build.
Morocco’s Europe-facing gateway, host to Oracle’s live Casablanca region.
Forty-plus cloud regions on why the constraint is the power and the capacity, not the equipment.
Why Singapore’s tropical liquid cooling standard fixed a doctrine gap, not a technology gap.
The Nordic advantage is real, and conditional, from a cooling supplier who declares his own bias.
Security designed as separate systems, and breached in the gaps between them.
One Nairobi campus and five numbers, read by someone who was there from the start.
In energy they want a megawatt. In compute they want an address.
Storage sized for sixty hours, not for a ride-through.
Energy companies are becoming compute enablers, not power suppliers.
Entelligencia’s interview series: the people building digital infrastructure, in conversation.