It was a shock when Iran's Islamic Revolutionary Guard Corps revealed during war games that they were using AI on home-grown drones, firing missiles that autonomously locked on to moving targets.
The leap was achieved despite advanced AI chips being officially off-limits to Iran due to sanctions.
It was one of several recent announcements indicating Tehran does not want to be left behind in the AI race. Under an official strategy, it aims to rank among the world’s top 10 AI powers by 2032, contributing 12 per cent to its gross domestic product.
Next month Tehran will host its first AI exhibition, featuring experts and training workshops. Iran’s goals raise a pivotal question: is this a blueprint for technological sovereignty, or a quixotic pursuit in a field dominated by resource-rich rivals such as the US, Saudi Arabia and the UAE?
"If Tehran is trying to build AI, it will be primarily to reinforce the security state, support asymmetric warfare in the region and project technological relevance despite isolation," said Mohammed Soliman, a technology expert and author of West Asia: A New American Grand Strategy in the Middle East.
But with foreign AI increasingly supporting the Farsi language, the fast pace of global tech could leave Iran sidelined. “Iranian models are impressive but small in scale; to match global competitors like GPT or Falcon Arabic, we need far greater computational power,” said Alireza Karami, an expert on industrial AI adoption.
“We end up relying on smuggled chips and second-tier solutions. That slows innovation, weakens reliability and leaves our projects far behind the global pace.”
Iran's ambitions
Iranian leaders have framed AI as a national priority. “AI is not just a technology, it is a new, shared language shaping the future of economy, politics, industry and human life,” said Hossein Afshin, Vice President for Science, Technology and Knowledge-based Economy.
A presidential-level AI organisation has been set up to oversee work across academia, industry and the public sector. While early statements spoke of investing $20 billion in AI development, only $115 million has been earmarked.
Iran has strengths in academic research. It ranks 15th globally and top in western Asia for AI research publications in the SCImago Institutions Rankings.
The research into AI spans fields such as smart agriculture, environmental modelling, health, energy and oil.
Then there is the military dimension that has brought the most tangible gains so far. The push to embed AI into defence systems accelerated after clashes with Israel beginning in 2024. Gen Kioumars Heydari, until recently the commander of Iran's ground forces, said in April that “AI is one of our priorities to maximise our defensive power".
According to IRGC‑linked sources, Iranian Ababil-5 and Mohajer-6 drones now carry missiles with AI-powered targeting, though specifics are not publicly available.
"The biggest practical leap has been in drones," military commentator Mehdi Bakhtiari told The National. He said Iran is also working to integrate AI into its missiles to boost their precision.
"AI now handles precision targeting, mid-flight course corrections, swarm co-ordination and even linking drones to missile systems for transmitting real-time target data," he said. "That’s where Iran has moved fastest, because drones are the backbone of our defence doctrine."
Sanctions and setbacks
Yet since AI began dominating headlines in 2022, it has become clear that simply seeking to make strides with AI does not necessarily make it so. Experts say that is especially true for Iran, which faces an uphill battle on several fronts to become a major AI player.
The computing power behind AI comes from chips known as GPUs, mainly produced by US tech giant Nvidia. With sales to Iranian companies blocked by sanctions, Iranian buyers have scoured grey markets and paid three times the global price to get their hands on the chips.
Global cloud services such as Amazon Web Services, Google Cloud and Microsoft Azure are also blocked for Iran. These vast server farms let companies train large AI models quickly and cost-effectively.
Aside from that, Iran's ageing energy grid and the persistent problem of electricity cuts will blunt even the most basic AI development efforts, said Mr Soliman, the technology expert, who is also senior fellow at the Middle East Institute.
"Modern AI is an energy story and Iran starts from a position of deep structural weakness," he said. "It has hydrocarbons but it doesn’t have the stable baseload power, modern grid, or cooling infrastructure required for large-scale data centres."
The sanctions, he said, stand in the way of several infrastructure needs presented by AI, such as cloud services and semiconductors, which can be difficult even for countries on friendlier terms with the US.
"Iran can smuggle small batches of legacy chips, but it will never be in the same league as Saudi Arabia or the UAE, which are securing Blackwell-class systems directly from the US, and that gap will only widen," he said.
Iran's research strength is also yet to produce many commercial wins for industry. Mr Karami said: “Iran produces a substantial body of AI research but the pathway from lab to market is still too narrow to carry it forward."
He pointed out that Iran last year ranked 91st in a Government AI Readiness Index, far behind the UAE and Saudi Arabia, which were both in the top five. The mismatch is particularly visible in sectors that could benefit quickly from applied AI, such as education, health and energy.
“Countries like Saudi Arabia and the UAE invest over $100 billion across AI sectors,” he said. “In comparison, Iran remains significantly behind.”
Reports suggest an increasing brain drain among Iranian AI graduates and young specialists, driven by limited demand. Without domestic experts, the sector would struggle to grow or compete with global rivals.
AI start-ups run by those who stayed behind are developing tools from language models to predictive analytics for health care and agriculture.
One notable example is Rakhsh AI, created by a small team of young developers. The language model chats in natural Farsi and can even create images from verbal prompts. It is designed to run smoothly on Iran’s domestic internet, avoiding reliance on blocked western services.
Potentially making matters even more challenging, Iran has a lengthy track record of cyber crime activity. A recent report from Microsoft showed that Iran seems to have shown no sign of slowing it down.
In August, seemingly annoyed by Iranian cyber attacks on the US, FBI assistant director Brett Leatherman said a hypothetical attack from Iran affecting US technology systems, data and infrastructure would be likely to be considered an act of war.
With that sort of well-documented rap sheet, all sanctions aside, companies in the private sector will likely think twice before trying to lend a helping hand to Iran, even in the context of AI.
