NVIDIA’s Bold US$1 Billion Move on Nokia Gives AI RAN the Credibility It Needs
By Larbi Belkhit |
06 Nov 2025 |
IN-7977
Log In to unlock this content.
You have x unlocks remaining.
This content falls outside of your subscription, but you may view up to five pieces of premium content outside of your subscription each month
You have x unlocks remaining.
By Larbi Belkhit |
06 Nov 2025 |
IN-7977
NEWSNVIDIA Invests US$1 Billion in Nokia and Announces AI-RAN Partnership |
During NVIDIA’s GTC Washington 2025, NVIDIA and Nokia announced a strategic partnership to add NVIDIA’s Artificial Intelligence (AI) Radio Access Network (RAN) technology to Nokia’s portfolio. Also, NVIDIA is set to invest US$1 billion in Nokia, taking a roughly 3% stake in the company.
T-Mobile US will also collaborate with the two companies to drive and test AI RAN technology as part of its 6G Research & Development (R&D) process, with trials expected to begin in 2026. During the press conference, Nokia commented that it expects early trials in 2026, first commercial deployments in 2027, and large-scale availability in 2028.
More specifically, Nokia will work to integrate NVIDIA’s latest AI RAN solution—the Aerial RAN Computer Pro (ARC-Pro). This solution leverages the NVIDIA Blackwell RTX PRO™ Graphics Processing Units (GPUs), consuming roughly 300 Watts (W) of system power in a telco-grade form factor, according to NVIDIA’s specifications.
Nokia will work to port its baseband software to the NVIDIA CUDA® platform, utilizing its anyRAN technology to enable flexible deployment. There will be two options for deployment: Commercial-off-the-Shelf (COTS), which will use the ARC-Pro platform in Dell’s PowerEdge servers, and an AirScale baseband where the GPU compute can supposedly be added in as an additional card.
IMPACTMarket Implications |
Today, the market sees little need to deploy GPUs at the cell site due to their unclear value proposition, cost, and power-hungry nature. The ARC-Pro appears to combine the best aspects of previous ARC generations—a significantly lower system power profile than the ARC-1, while delivering much stronger token generation capabilities than the ARC-Compact. It also comes with a more attractive price point and power profile.
One of NVIDIA’s weaknesses has been its channel to market. This new partnership with Nokia helps address this issue by providing a reliable, credible, and proven channel into the telco network. Given that the ARC platform will supposedly be backward-compatible with Nokia’s AirScale baseband, NVIDIA now has immediate access to a Serviceable Addressable Market (SAM) of Nokia’s entire deployed AirScale footprint globally. While power capacity and value proposition constraints will remain, a major barrier to entry has been removed by one of the industry’s leading vendors.
For Nokia, this partnership is an aggressive move to elevate the relevance and brand of its Mobile Networks unit. In 3Q 2025, the division saw 4% Year-over-Year (YoY) net sales growth, but also saw declines in gross and operating margins, at 34.8% and 0.7%, respectively. Last year, these metrics stood at 38.5% and 5.4%, respectively. This reflects the ongoing impact of being replaced by Ericsson in AT&T’s network and only having T-Mobile US as a customer in the highly profitable U.S. market.
With the portability of its anyRAN solution, Nokia now offers a more diversified compute strategy beyond its custom ReefShark System-on-Chip (SoC). The next step is porting its RAN software to run on NVIDIA’s CUDA® platform, which will likely be achieved by early 2026. While this is being undertaken, it is unlikely that Nokia will abandon its ReefShark program given the uncertain future for what is still the AI RAN concept, but likely investments may be scaled back to improve the overall profitability of the Mobile Networks unit.
From a macro perspective, T-Mobile US being the first Western brownfield operator to explore AI RAN formally will likely entice other operators to follow. Similarly, Nokia’s competitors will be working to port their software onto the CUDA® platform. For both groups, the fear of missing out on a fundamental shift in telco infrastructure will be the motivation. Ericsson has been advancing in this area, and Samsung signaled its intent earlier this year through an announcement with NVIDIA at Mobile World Congress (MWC).
RECOMMENDATIONSShort-Term Predictions |
NVIDIA will continue to improve the capabilities of its ARC platforms moving forward and will likely establish new partnerships with other incumbent infrastructure vendors (such as Ericsson), in a similar fashion to how it has done in the AI data center market with investments in OpenAI, CoreWeave, and others.
We have already seen a widespread deal announced with Samsung to acquire 50,000 NVIDIA GPUs for manufacturing and a commitment to work together, along with South Korean telcos and research institutions, on AI RAN development. ABI Research forecasts the following to occur over the next few months:
- Nokia will conduct significant market education discussions with customers on the long-term future of the ReefShark SoC and how they can upgrade existing cell sites to NVIDIA GPUs.
- ABI Research expects that Ericsson will accelerate its effort to partner with NVIDIA and announce its own partnership with NVIDIA and its ARC platform before MWC Barcelona 2026.
- NVIDIA will announce an update on its trials with T-Mobile US and Nokia halfway through 2026, as well as unveil a newer version of its ARC platform based on its Rubin GPU microarchitecture at a GTC event.
- Public statements from C-Suite executives regarding the feasibility of the AI RAN concept will become much less dismissive and operators will pursue more formal trials to explore the technology.
Written by Larbi Belkhit
- Competitive & Market Intelligence
- Executive & C-Suite
- Marketing
- Product Strategy
- Startup Leader & Founder
- Users & Implementers
Job Role
- Telco & Communications
- Hyperscalers
- Industrial & Manufacturing
- Semiconductor
- Supply Chain
- Industry & Trade Organizations
Industry
Services
Spotlights
5G, Cloud & Networks
- 5G Devices, Smartphones & Wearables
- 5G, 6G & Open RAN
- Cellular Standards & Intellectual Property Rights
- Cloud
- Enterprise Connectivity
- Space Technologies & Innovation
- Telco AI
AI & Robotics
Automotive
Bluetooth, Wi-Fi & Short Range Wireless
Cyber & Digital Security
- Citizen Digital Identity
- Digital Payment Technologies
- eSIM & SIM Solutions
- Quantum Safe Technologies
- Trusted Device Solutions