The Gulf Signed 47 GW of Solar. 14.8 GW Reached the Grid

Why the binding constraint on Gulf energy and AI data centre buildout in 2026 is neither capital nor chips but the queue for an energised grid connection.

AsiaPresswire – Two record buildouts are running across the Gulf at the same time in 2026, and they are colliding over the same scarce input. Saudi Arabia and the United Arab Emirates are signing renewable capacity at record prices and announcing data centre campuses by the gigawatt. Neither programme is short of money. Both are short of something much less glamorous: a substation bay, a transformer slot, and a date on which the thing actually gets energised.

Start with the number that sets up everything else. Saudi Arabia has a 2030 renewable target of roughly 130 GW. By early 2026 it had about 12.3 GW connected, rising to 14.8 GW by May 2026 against a year-end target of 20 GW. The Power Procurement Company had signed power purchase agreements covering more than 47.7 GW. Contracted capacity is roughly three times connected capacity, and the target is roughly nine times it. Closing that requires about 23 to 25 GW of new grid connection every year from 2026 to 2030. The Kingdom added 5.76 GW across all of 2025.

That arithmetic, as published in 2026, is the story. The rest of this piece is about which variable actually moves it, which two indicators let you watch it from outside, who loses first when it does not move, and what would prove the whole thesis wrong.

The structural variable is connection, not generation

It is tempting to read the gap between 47.7 GW signed and 14.8 GW connected as a construction problem. It is not primarily that. Successive tender rounds have produced some of the lowest solar tariffs ever recorded, including a levelised cost of US 1.56 cents per kilowatt hour at the Al Ghat wind project and a US$1.04 per MWh solar bid at Sudair. Generation is cheap and developers are willing. What is not available on demand is the equipment and approvals that stand between a finished solar farm and a wire that carries its output.

The specific bottlenecks, in the order they bite:

  • High-voltage transformers and gas-insulated switchgear, with global lead times of three to four years for large units as of 2026. Demand for generator step-up transformers grew 274 per cent between 2019 and 2025.
  • Interconnection studies, typically 24 to 72 months for large loads in 2026, and quoted at five to seven years in the most constrained markets.
  • Transmission buildout itself. Saudi Electricity Company has committed US$126 billion through 2030 for 380 kV substations and 12,000 kilometres of high-voltage lines.
  • Local content rules, currently 33 to 35 per cent and rising to 40 to 45 per cent after 2028, which push developers toward domestic manufacturing that is still scaling.

Grid investment alone is projected to account for close to 40 per cent of total power sector investment across the Middle East and North Africa over the next decade. When the grid share of capital spending approaches the generation share, the grid is the programme.

This is the structural variable, and it is measurable, which is what makes it useful rather than merely true. You cannot observe “execution risk”. You can observe quarterly megawatts connected against quarterly megawatts awarded, and you can observe transformer lead times.

Two indicators, and an honest answer about whether they are independent

The analysis type demands two leading indicators and an answer on whether they are independent. Here are the two, and the honest answer is that they are not, which turns out to be the point.

Table 1: The two leading indicators and their 2026 readings

Indicator What it measures 2026 reading Which side
Contracted-to-connected ratio Renewable delivery rate 47.7 GW signed vs 14.8 GW connected in May 2026 Supply
Announced-to-operational ratio Data centre conversion rate 467 MW operational in Q1 2026 vs a 1.5 GW national 2030 target Demand

 

Source: Azal Advisors 🔗, May 2026; NextFin 🔗, Q1 2026

The first tracks whether signed renewable capacity is turning into connected capacity. The second tracks whether announced data centre capacity is turning into operational capacity. They are published by different bodies, they describe different physical assets, and neither is derived from the other, so they are independent as measurements.

They are not independent as physics. A solar farm waiting on a 380 kV transformer is waiting in the same queue as a hyperscale campus waiting on a 380 kV transformer. The two indicators share one upstream dependency, and that shared dependency is exactly the structural variable identified above. Watching both is not redundant; it is the only way to see the constraint from two sides at once, and to check whether one side is winning the queue at the other’s expense.

Where the variable is heading, and why the Gulf’s advantage is borrowed

The Gulf currently holds a striking edge on connection speed. S&P Global puts grid connection at 12 to 18 months for data centres in Saudi Arabia and the UAE, against four to seven years in North America and Europe. Add electricity tariffs of US$0.05 to US$0.06 per kilowatt hour in the Gulf against US$0.09 to US$0.15 in the United States, and the region looks structurally advantaged on the one input that now binds.

That advantage is imported. Every transformer, most switchgear and a large share of high-voltage cable are made outside the region, in supply chains that the United States and Europe are also drawing on for their own grid programmes. The Gulf’s 12 to 18 months is therefore not a permanent feature of the region; it is a current position in a global queue. If transformer lead times stretch further, the Gulf advantage compresses even if nothing local changes.

Demand is rising into the same pipe at the same time. Grid-supplied electricity demand in Saudi Arabia reached 349 TWh in 2025, up 7.9 per cent year on year. Data centre capacity across the GCC is projected to roughly triple to more than 3 GW over five years, with Saudi Arabia and the UAE accounting for more than 80 per cent of additions, and average rack densities rose from 16 kW in 2025 to 27 kW in 2026. Higher rack density means each megawatt of IT load pulls harder on cooling and on the grid side, and a campus that needs 1 GW of IT load requires materially more than 1 GW of grid connection once cooling, losses and redundancy are counted.

Two things are therefore moving in the same direction at once: the volume of generation waiting to connect, and the volume of load waiting to connect. Both are growing faster than the connection rate.

The UAE half of the queue, which is a different shape

Treating the Gulf as one buildout hides a real divergence, and the two largest programmes are not running the same race.

The UAE entered this cycle as the regional renewables leader and has been overtaken on deployment speed by Saudi Arabia. Its anchor assets are large and already built: the Mohammed bin Rashid Al Maktoum Solar Park had 3,860 MW operational against a 2030 target of more than 8,000 MW, and the Barakah nuclear plant reached 5.6 GW at full operation in 2024. Solar accounted for around 12 per cent of UAE generation in 2024. That is a mature fleet expanding on a known grid as of 2026, in a country where DEWA in Dubai and EWEC and TAQA in Abu Dhabi run separate but interconnected systems, and where Dubai already counted 402 transmission substations by mid 2026.

The Emirati constraint is therefore less about connecting remote desert generation and more about concentrated load. The 5 GW UAE-US AI campus in Abu Dhabi, unveiled in 2025, is designed as the largest AI infrastructure project outside the United States, with a first phase around 200 MW targeted for 2026. Putting 5 GW of eventual load into one location is a transmission problem rather than a procurement problem, and it is being solved in a system that is already dense.

Saudi Arabia has the opposite profile: abundant land and record-cheap generation sitting far from load centres, and a transmission programme sized to close that distance at US$126 billion through 2030. One country is constrained by where the load lands, the other by how far the generation has to travel. Both end up waiting on the same transformers.

Storage is the pressure valve, and it is not yet big enough

Battery storage is the obvious answer to a connection-constrained system, because it lets generation that is already built deliver more of its output into a constrained window. In 2026 the scale is still small relative to the problem.

Saudi Arabia had about 8 GWh of battery storage installed in 2026, against a 22 GWh target for that year and a 48 GWh target for 2030, with roughly 30 GWh in the pipeline. A 2 GW, 7.8 GWh standalone system was synchronised in late 2025. Set that 48 GWh against a 2030 generation mix target of half gas and half renewables, and against 349 TWh of grid-supplied demand recorded in 2025, and storage is currently sized as a grid-services product rather than as a substitute for transmission.

That matters for the thesis because storage is often presented as the way around the queue. At 2026 scale it is not. It firming what is already connected, not a route to connecting what is not.

Who gets hit first

The scale of the announced pipeline is worth stating before discussing who loses, because it sets the size of the mismatch. The region is tracking more than 174 active and planned data centre projects with combined value above US$93 billion in 2026, and technology spending across the Middle East and North Africa is forecast to reach US$169 billion for that year. Both figures describe commitments. Neither describes energised load, and the distance between those two categories is where the losses will concentrate when the queue is allocated.

The queue is not first come, first served. It is allocated, and the allocation is visible if you look at who is building what.

The first group to lose is ordinary enterprise colocation. Gulf operators are absorbing sovereign AI campuses with state backing and multi-gigawatt ambitions. HUMAIN, the Public Investment Fund’s AI company, has an interim target of 1.9 GW by 2030, which runs about 27 per cent ahead of the Kingdom’s own national target of roughly 1.5 GW, and its longer pipeline points at 6.6 GW. A single planned 6 GW campus in Riyadh is being designed around a 380/132/33 kV network with two dedicated substations rated 500 MVA and 200 MVA and a 2,000 MVA bulk supply point. When a bank needing four racks and a sovereign campus needing a bulk supply point compete for the same substation capacity, the bank does not win the queue position.

The second group is merchant and second-tier renewable developers. Capacity with a sovereign off-taker and a PIF-linked sponsor will get its connection scheduled. Capacity without one will sit in the interconnection queue while the lead-time clock runs, and PPA tenors do not extend to accommodate a four-year transformer wait.

The third group is geographic. The GCC Interconnection Authority links Saudi Arabia, Kuwait, Bahrain, Qatar and the UAE today. Oman’s direct connection, a roughly 528 kilometre line from a 400 kV station at Ibri to a 400 kV station at Silaa in Abu Dhabi, is under construction and expected in the fourth quarter of 2026. Until it lands, Oman sits outside the shared reserve. Abu Dhabi Fund for Development and GCCIA have committed about AED 752 million, roughly US$205 million, to expand UAE interconnectivity, which tells you where the priority sits.

The fourth group is anyone sited for submarine cable latency rather than resilience. Drone strikes in March 2026 damaged Amazon Web Services facilities in the UAE and Bahrain, causing structural damage, disrupted power delivery and fire-suppression water damage that took the sites offline. The response has been site dispersion, blast-resistant construction and underground installation, which adds 5 to 7 per cent to capital cost. Coastal hubs optimised for latency are now being re-underwritten as security exposures.

What is not being disclosed, and why that matters more than the gap itself

The most useful finding in this analysis is not the delivery gap. It is that the region’s own capacity accounting cannot agree with itself.

Published estimates of Saudi operational renewable capacity in early 2026 range from about 8 GW to 14.8 GW. Independent projections for 2030 range from 45 GW to 90 GW. Third-party estimates of Saudi data centre capacity range from 222 MW on a colocation-only basis, to more than 410 MW, to 300 plus MW, depending on what is counted. When a market’s own capacity numbers disagree by close to a factor of two, a buyer cannot independently verify how much headroom sits behind a sales quote.

Table 2: Saudi renewable capacity, three published vintages for the same year

Measure Figure As of Source type
Connected renewable capacity 12.3 GW Early 2026 Sector analysis
Connected renewable capacity 14.8 GW, 17% of installed capacity May 2026 Sector analysis
Total renewable operational about 8 GW Q1 2026 Market research
Signed PPA capacity more than 47.7 GW Early 2026 Procurement reporting
2030 target 100 to 130 GW 2026 National target

 

Source: Azal Advisors 🔗, May 2026; Earth Energy Log 🔗, Q1 2026

Two consequences follow. First, any model built on a single capacity figure carries an error bar most modellers do not write down. Second, the gas hedge makes more sense than it looks. Saudi Arabia is deploying about 42 GW of new combined-cycle gas capacity, engineered to be carbon-capture ready, in parallel with the renewable fleet. Read as a hedge against the delivery gap rather than as a retreat from renewables, that number is the most honest admission in the whole programme: planners are building dispatchable capacity precisely because they cannot be certain when the renewable capacity will connect.

What would prove this wrong

Two signals, both falsifiable, both checkable from outside.

The first is a drainage signal, checked quarterly through 2027. If two consecutive quarters show grid-connected renewable additions exceeding new PPA awards, the pipeline is emptying faster than it fills and the constraint is easing. That has not happened; through 2025 and into 2026 awards have run far ahead of connections.

The second is an equipment signal. If transformer and switchgear lead times fall back below about 18 months while Gulf tender volumes hold or rise, the physical constraint is loosening and the 12 to 18 month connection promise becomes durable rather than positional. Current readings run 24 to 36 months as standard and up to four years for large high-voltage units.

Either signal would break the thesis. Neither is close today, which is why the thesis stands and why it has a shelf life.

If you are underwriting or siting in the Gulf this year

Five checks, in the order they will cost you money.

  • Get the interconnection date in the contract, not the commercial operation date. A campus that is built but not energised earns nothing, and the two dates can be years apart.
  • Ask what the connection capacity number measures: IT load, gross facility power, or contracted grid capacity. Published figures in this market differ by design, and the answer changes your economics.
  • Underwrite transformer and switchgear procurement as a critical path item with its own contingency, at three to four years for large units, not as a line inside the EPC schedule.
  • Check queue position against sovereign-backed projects in the same substation catchment. If a state AI campus is ahead of you, model the delay, not the announcement date.
  • Price security relocation explicitly. Dispersion and hardening run 5 to 7 per cent on capital cost, and coastal latency-optimised sites reprice as exposure after March 2026.

Do the first one this week. Ask your counterparty to state, in writing, the date on which the grid operator has committed to energise. If the answer is a commercial operations date rather than an energisation date, you have found the gap in your own project, and it is the same gap that sits between 47.7 and 14.8.

Data basis

Data as of October 2026. The primary metric is the ratio of contracted to grid-connected capacity, in gigawatts, taken from procurement reporting and sector analysis rather than from a single official series, because no single official series publishes both on the same basis. Electricity demand, tariff and equipment lead-time figures are from sector analysis and rating agency reporting. Data centre figures are operational IT or facility capacity as stated by each publisher; they are not reconciled to a common definition because the publishers do not use one. Where published estimates of the same quantity disagree, this piece reports the range rather than selecting a value. Currency figures are as published and not converted. No proprietary or terminal-sourced data was used.

Limitations

  • Saudi connected renewable capacity is published differently by different analysts, with early 2026 readings from about 8 GW to 14.8 GW; all derived ratios carry that spread.
  • Data centre capacity definitions differ by publisher, so the announced-to-operational ratio is indicative rather than exact.
  • No Gulf operator publishes confirmed interconnection points or energisation dates, so queue position cannot be observed directly, only inferred from project backing and equipment lead times.

Sources