Let me start by giving the Gulf its due, because the Western commentary rarely does.
Saudi Arabia's HUMAIN was launched in May 2025 by the Crown Prince, who also chairs it, sitting inside a Public Investment Fund (PIF) with close to a trillion dollars in assets. Its remit runs the entire stack, data centres, cloud, foundation models, applied products, all inside a single profit and loss rather than a portfolio of related companies. Tareq Amin, who built Rakuten Mobile's network in Japan, runs it, and his stated ambition for 2026 was to build the capacity Saudi built over the last twenty years in a single year. Eleven data centres under construction, 600,000 Nvidia GPUs planned, a 6.6 gigawatt pipeline.
The UAE moved earlier and differently. G42 and MGX, both orbiting Sheikh Tahnoon bin Zayed Al Nahyan, took the integrator path, assembling clouds, models and chips through connected entities and anchoring partnerships with US frontier labs. Stargate UAE with OpenAI, Oracle, NVIDIA and SoftBank targets a gigawatt cluster. MGX closed its first fund in July at forty-nine billion dollars, four billion past target, the largest dedicated AI fund ever raised. Qatar moves more quietly through its sovereign fund and Fanar, its homegrown model, and has deliberately chosen infrastructure without competing on foundation models.
This is not a region buying technology from other people. It is a region building the layers, including the ones the West assumed only the West would own.
LEAP, and the layer nobody listed
LEAP closed in Riyadh three days ago. Fifth edition, held under the theme Into New Worlds, four days at Malham, more than a thousand speakers, 1,800 technology brands, 600 startups and 1,900 investors. Around fifteen billion dollars of launches, investments and partnerships announced, on top of the 44.2 billion the ministry attributes to the first four editions.
The substance underneath the headline was real. AMD, Cisco and HUMAIN confirmed their first Instinct GPU systems are live and serving customers, with up to 250 megawatts planned from 2027 and a gigawatt by 2030. AWS confirmed its first Saudi cloud region for December 2026, and Microsoft its Azure region for November. Adobe and HUMAIN announced an image model built on Firefly Foundry designed for Saudi culture and Arabic prompts.
One analysis of the week listed what Saudi entities had assembled across the event. Power, data centres, compute, cloud, networks, applications, manufacturing, capital.
Read that list again and notice what is missing.
Every layer required to produce artificial intelligence was represented. Not one layer required to monetise it. There was no track on pricing architecture, no keynote on how you charge for an outcome rather than a token, no session on what compensation looks like when the product does the work. There were dancing robots and a brain-computer interface demonstration.
The stated ambition was democratisation, getting this capability to the masses, to local markets and to global ones. It is a good ambition and I have no argument with it. However democratisation is a distribution goal, not a commercial model, and the two are constantly confused.
We have seen this sequence twice before. The industrial revolution built the machines decades before anyone worked out the firm structures, pricing and labour models to make them pay, and the early industrialists who won were rarely the ones with the best machinery. The internet democratised distribution and information almost overnight, then spent fifteen years failing to work out how to charge for any of it, which is why an entire generation of infrastructure was built by companies that captured almost none of the value created on top of it.
The technology moves at the speed of capital. The business model moves at the speed of institutional courage, which is considerably slower, and that gap is where the value goes to somebody else.
However.
Every announcement I have read from the Gulf in the past eighteen months describes gigawatts, GPUs, land plots and model parameters. I have not seen one that describes how any of it will be monetised.
The mistake, repeated at speed
The West made a specific and expensive error. It rebuilt the product and left the commercial operating model exactly as it was. Pricing designed for seats, compensation designed for a single moment of closing, retention metrics that count billing rather than value, and a GTM motion aimed at a buyer who no longer behaves that way. Two years and several trillion dollars of market value later, that error is now visible in the share prices of companies that shipped genuinely excellent AI.
The Gulf is running the same play with more money and less legacy.
More money is obvious. Less legacy should be an advantage, and this is the part that frustrates me most. HUMAIN has no decades of seat based licence revenue to protect. No installed base conditioned to per-user pricing. No sales organisation whose compensation plan has calcified around bookings. It could design a commercial operating model for agentic delivery from a blank sheet, which is a position no incumbent on earth currently enjoys.
Instead the region is importing the Western model wholesale. Sovereign compute sold like cloud. Enterprise AI sold like enterprise software. National champions building world-class infrastructure and then reaching for a pricing page that was obsolete before the first rack was energised.
Here is the sharpest version of the claim, and I would defend it in any room in Riyadh, Abu Dhabi or Doha.
Sovereign capital can buy every layer of the stack. It cannot buy a commercial operating model. That has to be designed, and nobody in the region has published evidence that anyone is designing one.
What that actually costs
Three consequences, and none of them are theoretical.
The first is margin. Compute is a capital intensive, depreciating asset with a brutal cost curve. Sell it on Western cloud logic, per unit consumed, and you inherit Western cloud economics without Western scale advantages. Amin has been explicit that a credible sovereign alternative is the strongest discipline on hyperscaler pricing, which is true and also a warning. If your commercial strategy is to be the cheaper option, you have entered a price war against companies whose marginal cost is falling faster than yours.
The second is the value gap. HUMAIN reports that its own payroll team is now one person, with agents handling the rest. That is a genuine, measurable outcome, exactly the kind of proof the entire market is desperate for. Priced as compute consumed, that outcome is worth a few dollars of tokens. Priced as value delivered, it is worth a great deal more. The region is currently metering the former and giving away the latter.
The third is the one that decides the race. Whoever monetises the layer above infrastructure captures the value of everything beneath it. That is the lesson of the last technology cycle and the Gulf lived it first-hand, because the region has spent two decades building world-class telecommunications infrastructure and watching the returns accrue to the applications riding over it. A pipe that delivers data instead of water is still a utility, however advanced the pipe.
The uncomfortable question for every sovereign fund reading this. Are you building an AI champion, or are you building a very expensive utility for someone else's business model?
Who is ahead, and who is exposed
Since the question is now openly a race, let me score it honestly.
The UAE is ahead on execution. G42 got there first, secured the US alignment that unlocks chip supply, and built an operating company that actually runs infrastructure at scale. In this cycle, early and aligned has mattered more than marginally larger.
Saudi Arabia is ahead on capital and structure. The single profit and loss is a genuine strategic advantage that almost nobody has noticed. Infrastructure, models and applications inside one balance sheet means HUMAIN could design one coherent commercial model across the whole stack, from kilowatt to outcome. Nobody else has that option. G42 portfolio structure makes it considerably harder.
Qatar is ahead on discipline, and I mean that as a compliment. Choosing infrastructure and deliberately not competing on foundation models is a clearer strategic choice than most Western boards have managed.
And all three are exposed on the same thing. Not one of them has articulated how the value gets captured. The gigawatts are announced in public, the commercial model is announced nowhere, and in my experience the second usually means it does not exist yet.
Who wins
The winner of this race will not be whoever deploys the most gigawatts, and it will not be decided by chip allocations from Washington.
It will be whoever first builds a commercial operating model designed for what agentic AI actually delivers. Pricing that captures outcomes rather than consumption. Compensation that rewards realised value rather than signed contracts. Measurement that proves what was delivered rather than what was billed. An organisation shaped around a continuous value loop rather than a sales cycle imported from 2015.
That is a design problem, not a capital problem, which is precisely why the region's greatest strength does not help with it. You cannot outspend it. You can only out think it, and the window to do that is open right now because the Western incumbents are trapped inside models they cannot unwind without destroying their own revenue base.
The Gulf has no such constraint. It is the only serious player in the world building this from a blank sheet, with the capital to execute and no legacy to protect. If HUMAIN, G42 or Qatar's sovereign vehicles get the commercial model right, they will not merely participate in this market. They will define how it is monetised, and everyone else will sell against their scoreboard.
If they do not, they will have built the finest infrastructure in the world and handed the value above it to whoever did.
The technology race is broadly settled. The stack is being built and the Gulf is building it. The commercial model race has not started, and it is entirely up for grabs.
Twenty years ago the region built the pipes and watched the value accrue elsewhere. The industrial revolution and the internet both ran the same script before that. The machines arrive first, the business model arrives late, and whoever closes that gap fastest keeps the value.
That mistake is available again, at considerably greater expense.
First published in the Agentic Commercial Model newsletter on LinkedIn, September 6, 2026. Read the original on LinkedIn.