SambaNova to invest $140 million in Saudi AI infrastructure & services
US AI firm partners with STC AI and MCIT to advance sovereign AI in Saudi Arabia
#SaudiArabia #ArtificialIntelligence – Palo Alto-based AI company SambaNova Systems has announced $140 million in investments in Saudi Arabia's AI infrastructure, partnering with the Ministry of Communications and Information Technology (MCIT) and the RDIA’s National Semiconductor Hub. SambaNova CEO Rodrigo Liang announced that the company is also launching a Sovereign LLM-as-a-Service in collaboration with STC AI, offering the world's largest open source model LAMA 405B and running other capabilities such as STC’s Enterprise GPT. Lastly, Liang announced that SambaNova will soon offer the full DeepSeek 671 billion parameter R1 model on SambaNova Cloud. The model will be made available to select customers first, before being made more widely accessible.
SO WHAT? – With a number of high-level partnerships with key public and state-backed organisations, SambaNova has been growing its business in Saudi Arbia over the past year. This investment announcement marks a significant step in SambaNova’s Saudi ambitions and also a further addition to the Kingdom’s fast-growing capacity to provide powerful compute resources in the country.
Here are some key points from the announcement:
SambaNova Systems, a leading US-based AI infrastructure company, is investing $140 million into Saudi Arabia’s AI ecosystem. The investments to advance AI and semiconductor capabilities in the Kingdom, include those made in partnership and collaboration with the Ministry of Communications and Information Technology (MCIT) and the RDIA’s National Semiconductor Hub.
The announcement was made by SambaNova CEO Rodrigo Liang, speaking on the main stage of LEAP25.
SambaNova and STC AI are launching a Sovereign LLM-as-a-Service cloud, featuring Meta’s LLaMA 405B, the world’s largest open-source AI model.
The Sovereign cloud service will include capabilities such as STC’s Enterprise GPT and provide industry-leading inference speeds.
In addition, SambaNova will soon offer the full DeepSeek 671 billion parameter R1 model SambaNova Cloud. This will be the full V4 model, not a derivative.
SambaNova also claims a major efficiency breakthrough, running the massive model on a single 10-kilowatt rack, significantly reducing infrastructure requirements from competitors technology that requires 40 racks.
Select customers will be able to access the DeepSeek model immediately, with broader cloud availability planned in the coming weeks.
ZOOM OUT – This week’s LEAP 2025 IT exhibition and conference showcases both progress in Saudi Arabia and investment in the country from global technology leaders. During the past year, the Kingdom has brought new focus to specific AI compute requirements partnering with a growing number of leaders in semiconductor development and cloud AI services.