Let’s check out what Pallavi Mahajan, Chief Expertise and AI Officer at Nokia, has stated about synthetic intelligence (AI) and the alternatives that AI-native networks and 6G create. Pallavi leads Nokia Bell Labs, Expertise and AI Management, and Group Safety to drive innovation in core applied sciences and strengthen AI and safety capabilities, in keeping with the corporate. Previous to becoming a member of Nokia, Pallavi held the place of Company Vice President (CVP) and Basic Supervisor (GM) of Knowledge Middle and Synthetic Intelligence Software program at Intel, the place she outlined and led the corporate’s AI software program technique.
Additionally Learn: What They Mentioned: Invoice Gates of Gates Basis on AI
What’s the position of networking in AI?
Based on Nokia, communications networks are crucial drivers of synthetic intelligence (AI) innovation and success. They unlock the potential of AI purposes by enabling seamless information transmission between units, cloud platforms, information facilities, and edge infrastructure.
As industries akin to manufacturing, training, finance, healthcare, transport and authorities embrace AI-driven mission-critical companies, they want strong connectivity to ensure these companies can operate effectively.
This creates vital alternatives for service suppliers, Nokia stated. By constructing an AI-optimized community infrastructure, you possibly can develop new revenue-generating companies, akin to community slicing, that cater to particular business wants.
Nokia says that service suppliers are turning to AI to simplify community operations, optimize efficiency, enhance safety, scale back vitality consumption, and predict and forestall downtime.
AI is a key enabler of autonomous networks, as envisioned and outlined by TM Discussion board, which may self-configure, self-optimize and self-heal with minimal human intervention. On the highest ranges of autonomy, AI performs an important position by enabling the community to handle complicated operations, adapt to adjustments and guarantee steady efficiency optimization all by itself.
Nokia introduced in November this 12 months a reorganization of its enterprise into two major working segments, which can take impact on January 1, 2026. Extra on this may be learn right here.
Additionally Learn: Nokia Unveils Strategic Overhaul to Lead AI-Pushed Community Period and Seize Worth From AI Supercycle
What They Mentioned: Timeline
November 2025:
Capital Markets Day 2025
Throughout Nokia’s Capital Markets Day 2025 webcast, Pallavi shared insights on AI-native networks. She stated: “As a technologist, I’ve a robust eye of recognizing when the subsequent wave of disruption goes to hit, journey that wave after which be the disruptor. I used to be born doing networking. After which a decade again, I moved to return and construct the compute for the AI infrastructure, began with the cloud, then to the sting after which to the information heart.”
“Now as I’ve been constructing the AI infrastructure, I can now see that the subsequent wave of disruption is coming. And this time, it’s going to disrupt the networks. Essentially, should you have a look at it, the AI supercycle, it has put an exponential progress on the calls for of those networks. Now whether or not it’s latency, whether or not it’s capability, whether or not it is bandwidth, whether or not it is reliability. And these metrics, they’ve at all times been very important. However at this time, they’re more and more turning into crucial in scale and in urgency as a result of this calls for a brand new type of community, networks that join intelligence and networks that turn into smarter through the use of that intelligence. And Nokia is the one Western firm which has a complete portfolio, spanning radio, core, entry, transport, IP, all the best way as much as the cloud, uniquely positioned to form and ship the networks of at this time and tomorrow. And this is the reason I am so excited to have joined Nokia because the Chief Expertise and AI Officer.”
From Cloud to Agentic AI
Explaining the evolutions of networks towards cloud and AI, she stated:
“ChatGPT, I feel it began with about 117 million parameters. And at this time, it is near about 2 trillion parameters. And at this time, we’ve entered the world of agentic AI. That is the place clever brokers work together, and there is huge machine-to-machine chatter. Now should you see, we have had huge expertise disruptions in such a brief span of time. And the important thing to this transformation was truly fueled by open ecosystems, open information, open framework, open fashions as a result of openness fosters innovation and collaboration.”
Bodily AI: The place Milliseconds Imply Security
“Now all through this AI supercycle, networks have been foundational and networks have additionally gone about and altered, whether or not you are speaking about scaling up for machine studying, whether or not we’re speaking about scaling out for generative AI or we’re speaking about scaling throughout for agentic AI. However now there may be one other disruption ready to occur on this AI supercycle. And I can truly see it coming as a result of AI is now on the cusp of the subsequent wave of transformation, which is bodily AI. That is the place the boundaries of the bodily and the digital worlds are going to get blurred,” she continued.
“Now give it some thought. That is the autonomous automobiles, drones, AR/VR glasses, clever factories, well being care. That is the world the place crucial important companies will demand that the community ought to at all times be on, the place subsplit second choices should be made and each millisecond issues. And with bodily AI, we may have robots standing hand-to-hand with people, with heavy equipment. And now simply think about that these robots, the security and the movement management loops that they want, they want choices to be taken within the matter of microseconds. Now one community slip and the robotic can truly miss a security cease. So in these environments, it is about security.”
“Gen AI was 3 years again. 2025 is the 12 months of agentic AI. And now I can clearly see that bodily AI is knocking at our doorways. And this evolution of AI will exponentially change the size of the community themselves. Now whether or not it’s bandwidth, whether or not it’s capability, whether or not it’s latency, whether or not it’s reliability, all of those KPIs are altering with AI.”
Showcasing a demo instance, she stated:
“The site visitors is bursty, the site visitors is unpredictable. And what you additionally see in right here is that the uplink site visitors is increased than the downlink site visitors. Actually, my groups in Bell Labs are predicting that within the new world, at this time, the downlink to uplink ratio is 12:1, and it’ll transfer to 4:1. And we’re seeing increasingly more AI-native site visitors originating from cell units, 48 % of ChatGPT, 61 % of Gemini’s AI site visitors is originating from cell units.”
“What we simply noticed was simply 1 AR expertise on one system. Tomorrow, the related panorama will look radically completely different. AR glasses are going to be frequent, as frequent as smartphones. We shall be utilizing them for navigation, for purchasing, for translation. 50 % of autonomous automobiles shall be driving by themselves. Tens of 1000’s of robots shall be delivering parcels. AI brokers shall be there in each retailer for pricing, for merchandising, for higher buyer engagement. After which take into consideration all town edge AI factories that shall be orchestrating site visitors, vitality, logistics, all in actual time.”
“Right this moment’s networks, they ship 99.999%. Now what does 99.999% imply? It signifies that the networks — there could possibly be minutes of downtime in a 12 months. However we have been speaking about robots on a manufacturing unit. When robots on a manufacturing unit flooring are working hand-in-hand with people, the downtime has to shrink to seconds. And we are actually beginning to discuss 99.9999%. 99.9999% reliability as a result of one missed security sign may cause an accident.”
“Now let us take a look at bandwidth. We talked about – we confirmed you an AR/VR instance. We’ve a lot of AI inferencing that’s taking place. And all of this – due to this, the bandwidth has already jumped 7x to 400 gigs per port. However tomorrow’s community, they want terabit class capability to feed these large AI fashions and real-time video streams. Now take into consideration all of the AR glasses which can be going to render 3D environments, autonomous automobiles, all of them streaming sensor information on the similar time.”
“And now let us take a look at the third metric, which is latency. Right this moment, latency is about 10 to twenty milliseconds, and that works properly for video calls. However think about, as soon as once more, the robots in a manufacturing unit flooring, when they’re making security choices, we’re speaking of sub-millisecond latency. And once you have a look at these numbers, you would possibly really feel like, okay, these are some small jumps that we’re speaking about. However as a technologist, having finished this in my life, I can inform you that with the intention to make these leaps, it’s plenty of exhausting work and plenty of analysis and growth that goes with the intention to make it occur. As a result of when the business moved from 9.99% to 99.999%, it took us plenty of work. And now we’re pushing even additional to get these networks prepared for AI-native site visitors.”
Talking in regards to the core community, which is cloud-native and hardware-independent, she stated the workforce is making it extra AI-native:
“Workforce is engaged on making it extra AI native as a result of on the earth of AI purposes, on the earth of AR units, on the earth of robots, they don’t seem to be going to talk 3GPP. So we’re evolving the core past the standard boundaries, making it programmable, making it agentic, making it service conscious, making it prepared for AI-native features, features like real-time translation, fraud detection and way more.”
She added:
“The KPIs that the networks of the long run demand could be very exhausting. And that’s the reason myself, because the Chief Expertise and AI Officer, I am hyper-focused on taking all of Nokia’s differentiating property and doing the exhausting factor, which is harnessing the wealthy information, the deep insights and the area experience that we’ve constructed over years and bringing AI into the protocols and the algorithms that made these networks occur. Now in brief, what I am speaking about is that the networks which have powered AI are actually going to turn into extra clever through the use of AI.”
Constructing Autonomous, Anticipatory Networks with AI
Introducing WaveSuite, a software program utility layer, she stated:
“Now we’re beginning to discuss networks that simply don’t react. These are networks that are actually anticipating. Actually, in a stay trial with du within the UAE, we noticed that through the use of this, the optimized efficiency modeling, it cuts the planning time in half, and it truly improved the design efficacy by 30%.”
She concluded:
“Our pursuit is easy. We are going to use each ounce of intelligence out there to us, our property, our information, our area experience and our partnership, our deep partnership with telcos and hyperscalers to construct networks for the brand new period of AI-native site visitors, networks that simply do not carry intelligence, however networks that repeatedly be taught, adapt, shield and enhance. And as we construct these networks, we’ll companion with the best-in-breed companions.”
“We are going to companion with of us in silicon, with of us in software program, with of us in platforms to assist our prospects unlock their subsequent wave of progress.”
Additionally Learn: What They Mentioned: Puneet Chandok of Microsoft on AI
December 2025:
New analysis commissioned by Nokia, surveying greater than 2,000 expertise and enterprise decision-makers within the U.S. and Europe, reveals sturdy demand for AI options – and a transparent business consensus that connectivity and community infrastructure should evolve to assist the subsequent wave of AI progress, Nokia stated in a press release on December 16, 2025.
Whereas the U.S. continues to steer world AI deployment and mass-market adoption, 88% of U.S. respondents expressed concern that the enlargement of community infrastructure could not hold tempo with AI funding. In Europe, 86% of enterprise respondents stated present networks are usually not but geared up to deal with widespread AI adoption. Two-thirds of these surveyed stated they have already got AI in stay use, and greater than half have already skilled challenges akin to downtime, latency, and throughput constraints related to growing information calls for.
“The primary wave of the AI supercycle has already reshaped industries and accelerated innovation,” stated Mahajan whereas releasing the brand new examine by Nokia. “This analysis reveals a transparent understanding throughout the ecosystem that future waves will demand extra superior, AI-native networks and substantial funding to strengthen community necessities. Connectivity, capability, and low-latency efficiency have gotten ever extra important components for remodeling how units work together, industries function, and folks stay and expertise expertise as AI strikes ahead.”
The analysis contains two studies: Infrastructure First Is the New America First (U.S.) and AI Is Too Huge for the European Web (Europe).
In an article titled Advancing Connectivity within the AI Period, dated December 16, 2025, Mahajan wrote: “The AI supercycle has begun, and very like the Web revolution, it’s essentially reshaping the position of connectivity and the way we use and expertise expertise. It’s also altering the site visitors patterns in telecommunications networks.”
“At Nokia, we imagine that safe, superior connectivity infrastructure is an integral enabler of AI innovation, connecting intelligence the world over’s units, the community edge and information facilities. We’re collaborating carefully with our strategic companions and prospects to drive the convergence of AI and radio entry networks, capturing the advantages of AI.”
She reiterated that 88% of U.S. telecommunications suppliers and enterprises and 78% of European respondents see connectivity and infrastructure as the largest boundaries to scaling AI.
AI-RAN, Nvidia, and the Highway to 6G
In October 2025, Nvidia and Nokia introduced a strategic partnership so as to add Nvidia-powered AI-RAN merchandise to Nokia’s RAN portfolio. Nvidia additionally introduced a USD 1 billion funding in Nokia to speed up AI-RAN innovation and the transition from 5G to 6G.
T-Cell U.S. is working with Nokia and Nvidia to combine AI-RAN applied sciences into its 6G growth course of. Trials are anticipated to start in 2026, targeted on subject validation of efficiency and effectivity beneficial properties for purchasers. The transfer will allow huge enhancements in efficiency and effectivity, serving to be sure that customers utilizing generative, agentic and bodily AI purposes on their units may have seamless community experiences. It should additionally assist future AI-native units, akin to drones or augmented- and virtual-reality glasses whereas being prepared for 6G purposes akin to built-in sensing and communications, Nokia stated.
Additionally Learn: What They Mentioned: Jamie Dimon of JPMorganChase on AI
Commenting on the partnership (Nokia and Nvidia), Mahajan stated Nokia is bringing the AI-RAN imaginative and prescient to life: “Our AI-RAN resolution, powered by the Nvidia ARC-Professional computing platform and Nokia anyRAN software program, permits RAN and AI workloads to run on a shared, GPU-accelerated platform, supporting software-defined RAN evolution,” she stated.
“Bodily AI programs, akin to robotic AI, autonomous automobiles, and superior units that allow immersive experiences, will demand ultra-low latency and big uplink capability. Networks might want to reply in milliseconds and within the subsequent part, even lower than one millisecond, to allow real-time communications,” she highlighted, noting that “Nokia has a key position in growing AI-native 6G structure, which shall be essential for making these capabilities a actuality.”
As soon as-in-a-Era Alternative for Connectivity
Mahajan concluded: “The AI supercycle is a once-in-a-generation alternative. To make sure that connectivity doesn’t turn into the bottleneck for AI innovation, we’d like shut collaboration and co-innovation throughout the business ecosystem.”
It is a growing story, and extra quotes and insights from Pallavi Mahajan shall be added as they turn into out there.
