Snapdragon Summit Day 3 – Tales from the Kingdom

Humain tells only half of the story.

• Humain has a strategy to reimagine the way that enterprise IT works, but did not say much about the real story, which is turning abundant real estate and energy in Saudi Arabia into compute for export that helps diversify the economy away from petrochemicals.
• Humain is the result of a re-organisation of several AI-related assets and companies within the Public Investment Fund (PIF, the sovereign wealth fund of Saudi Arabia) into a company that is focused on developing AI in the kingdom as well as creating a thriving export business.
  • First, AI for enterprise: which is a complete reimagining of how IT inside a company runs.
  • Humain has taken a leaf out of Amazon’s book and is testing it on itself before offering it more widely to the rest of the world.
  • The idea is to have a series of agents, each specialised in a certain task, which are then activated from a single prompt where the user simply tells the system what he or she wants to do.
  • The key to this is a multi-orchestration layer that the company has built that is capable of understanding the request and activating the right agent to carry out the task.
  • These tasks are all corporate-related, such as making payroll to booking sick leave, and the company says it has built 120 agents to tackle enterprise IT tasks.
  • However, only a few of these are available at the moment and include booking leave and making payroll, as when a user asked the task agent to task the CEO to give them a pay rise, the system amusingly refused to do it.
  • This is a bold proposition as Humain is basically claiming that Salesforce, Snowflake and SAP are obsolete and can effectively be replaced with agents and an orchestration layer.
  • This is going to be a tough pill to swallow for companies that have already made large investments in these areas, who I suspect would prefer to simply put an agent on top to control these services.
  • This means that this idea will appeal most to small businesses and start-ups that have no legacy and as an emerging economy, Saudi Arabia has plenty of these sorts of companies.
  • This went hand in hand with the launch of a custom laptop (Humain Pro), which has been designed in house and has the orchestration layer and Arabic LLM (ALLAM) already integrated into the device.
  • This is a Qualcomm-powered device and is one way that Humain intends to distribute the enterprise offering to its clients.
  • Second, AI compute: which I think is where the real opportunity lies and what I suspect interests the Kingdom the most.
  • Saudi Arabia has abundant real estate and energy, which are two commodities that are in short supply in many other regions of the world.
  • This, combined with a novel approach to inference and chipset selection, leads Humain to think that it can offer inference compute at a substantial discount to everyone else.
  • This appears to have some traction as the company claims it has already sold 1GW of compute before the data centres come online.
  • The company will be pitching foreign entities to run their compute in Saudi Arabian data centres, which will count as exports from Saudi Arabia.
  • This is crucial because if it becomes a large business, it will start to have an impact on the long-term strategy of diversifying the Saudi Arabian economy away from petrochemicals.
• The net result is that there are two pieces to the Humain story, which are a bold plan to reinvent enterprise IT processes and to create a thriving export business to diversify the economy.
• I suspect that it is in this second business that Qualcomm is most interested, as it has all but admitted that there is a data centre product in the works for which Humain could be the first customer.
• Furthermore, there have also been discussions with one of the hyperscalers, which would not have occurred if Qualcomm had nothing more than slides to show.
• Humain says it has already sold 1GW of compute, and there is probably a lot more to come, meaning that Humain is a significant opportunity from a revenue perspective for Qualcomm.
• It would also represent a crucial first customer from which it will become much easier to win others once one can point to a real data centre and show the benefits that have been delivered.
• This is another potential revenue stream, like PCs, where I think the forecasts from the market are too low, meaning that they could provide growth that the market has not anticipated.
• Overall, Snapdragon Summit 2025 has delivered the building blocks to ensure that Qualcomm will continue to increase its position in markets outside of smartphones and to be ready for AI when it comes to the edge.
• Hence, market estimates are probably still too low over the next 5 years, meaning that the shares are actually cheaper than the headline calculations would dictate.
• Hence, I remain very comfortable with a position in Qualcomm and will look to buy more the next time the market decides to send the shares down 10% for no reason after it reports good results.