Imagine a network that can think, adapt, and heal itself — all without human intervention. Autonomous networks are the future of telecom: intelligent, self-managing, and capable of delivering seamless, reliable connectivity at unprecedented scale.

We’re looking for bold, AI-driven solutions that make networks truly autonomous — from real-time adaptation and self-healing to intent-based orchestration and edge intelligence. If you’re pushing the boundaries of what networks can do on their own, this is your stage.

What We’re Looking For

Here are some priority areas, though we welcome bold, creative ideas beyond this list:

• Self Healing Networks: How could autonomous systems, detect, diagnose and resolve network faults incl. Hardware failures or even degraded performance without human intervention?
• Autonomous 5G slice adaptation: How would AI driven network auto provision, scale and adjust network slices based on real time usage patterns & service demands?
• Intent based networking: How might networks interpret high level business or user intents and translate them into autonomous, adaptive network configurations without manual interventions
• Dynamic Resource Sharing: How might autonomous systems allocate spectrum, compute and transport resources in real time across multiple services or tenants?
• Edge Autonomy: How might networks autonomously decide whether to process data at the edge vs cloud depending on latency, load balancing or privacy concerns and dynamically shift functions as required?
• Autonomous Planning: How can networks autonomously plan, test & validate new infrastructure rollout with minimal or no human intervention?
• Tool-augmented AI Agents: How can autonomous networks leverage tool-augmented AI agents to understand operator intent with the objective of implementing closed loop automated systems that optimally configure networks while maintaining policy and security constraints?

Why It Matters

Autonomous networks are the future of telecom — delivering faster adaptation, greater reliability, and smarter resource use while reducing complexity and costs. We’re looking for scalable ideas that can be tested, proven, and applied widely. Your innovation could shape the next era of intelligent connectivity.

​

Are you driving innovation in telecom energy efficiency? We’re looking for groundbreaking, scalable solutions that will help us create a more sustainable and energy-conscious future for telecommunications.

This year’s challenge is aligned with our vision of building the AI-native telco — autonomous, customer-focused, and intent-based. With a special focus on Energy Efficiency & Zero Bit Zero Watt, we seek solutions that minimize energy consumption across the entire network lifecycle—from design to daily operations—using the power of artificial intelligence (AI). The goal isn’t just to save energy, but to do so at scale, across millions of network components worldwide.

The Challenge

The telecom industry faces a dual responsibility: meeting ever-growing data demands while reducing its environmental impact. Our ultimate vision is a network that consumes energy only when data is actively being transmitted—what we call “Zero Bit Zero Watt.”

Achieving this requires fresh thinking and AI-driven solutions that deliver measurable, tangible energy savings. AI and machine learning can move us beyond static, scheduled power-saving modes toward intelligent, real-time systems that adapt dynamically to network traffic. From enabling deep-sleep states for underutilized components to optimizing routing for energy efficiency, we want to explore ideas that span the entire telecom stack—from RAN to transport to core networks.

The key: solutions that are both innovative and practical to scale, easily integrated into existing networks, and deployable across multiple countries.

What We’re Looking For

We are seeking AI-powered innovations that can demonstrate clear impact, scale, and feasibility. Strong proposals will:

• Reduce Network Power Consumption at Scale: Offer new methods to cut energy use in both active network elements and passive infrastructure, such as cooling systems, using e.g. AI-driven insights that can be applied consistently across thousands of sites.
• Improve Operational Efficiency: Use AI and automation to optimize energy use in real time, adapting seamlessly to traffic patterns, loads, and network conditions.
• Enable Smarter Network Design: Introduce AI-powered design principles or tools that make networks more energy-efficient from the ground up, including approaches like green coding and green AI.
• Provide a Clear Path to Implementation: Demonstrate vendor-agnostic solutions that integrate easily without large-scale hardware overhauls, ensuring rapid adoption across diverse networks.
• Maintain Customer Experience: Ensure that efficiency gains do not come at the expense of performance or reliability, with AI helping to balance energy savings and quality of service.
• Demonstrate Alignment with Standards: Solutions should follow international standards (e.g., 3GPP, O-RAN, ITU) to ensure interoperability.
• Show Measurable Impact: Define clear KPIs such as percentage energy savings, quality-of-service metrics (e.g., call drop rates, throughput stability), and TCO reduction.
• Outline integrability: Solutions must be capable of being integrated into a given legacy landscape (brownfield). Any efforts for such an exercise must at least be indicated. A solution that works over-the-top would need to indicate how it interoperates with the given infrastructure, etc. APIs and interfaces must be declared.
• Grids: How might we use AI to support micro-grid or dynamic grid-interactive (e.g. eco-district) solutions with data center and network workloads?

Example Areas of Interest

Here are some priority areas, though we welcome bold, creative ideas beyond this list:

• Network-wide AI Platforms: Unified energy management across multiple network layers and vendors through a central AI platform.
• Predictive Maintenance & Power Optimization: AI models that forecast hardware failures, optimize maintenance, and dynamically tune power settings based on real-time traffic.
• AI for Zero Bit Zero Watt: Cutting-edge AI approaches that bring us closer to near-zero power consumption during idle times, with rapid wake-up capabilities.
• AI for Facility Management: Large-scale energy optimization for data centers, central offices, and related infrastructure.
• Green Coding & Green AI: More efficient algorithms and models that minimize the energy footprint of AI itself, enabling sustainable large-scale deployment.
• Lifecycle Sustainability and Circular Economy: Extending energy efficiency to equipment manufacturing, reuse, and recycling, reducing the total carbon footprint across the full lifecycle of telecom networks.

Why It Matters

Your application should clearly demonstrate the impact on energy savings, operational efficiency, and scalability. Practicality and the potential for broad implementation will be key differentiators.

We look forward to your innovative ideas that will shape a more sustainable, energy-efficient future for telecom.

​

In a world continuously challenged by crises—from geopolitical tensions and climate disasters to economic uncertainties—building resilient supply chains is more crucial than ever. To proactively address these challenges, we seek innovative companies with an AI-first mindset that leverage artificial intelligence to make our supply chains more transparent, robust, and adaptive.

Potential solutions span from deep supplier network analysis and early risk detection to automated sourcing alternatives and sustainable refurbishment strategies. AI offers new ways to uncover hidden risks early, identify alternative supply options, and optimize planning to minimize disruptions while controlling costs.

Key questions we want to explore include:

• Supply Chain Transparency: How can AI increase visibility across the entire supply chain, including suppliers of suppliers, and beyond, to uncover hidden risks related to critical suppliers?
• Early Warnings: How can AI provide early warning systems and impact assessments for supply risks caused by trade conflicts, price volatility, climate events, geopolitical shifts, or acquisitions?
• Supply Mitigation: How can AI help identify alternative sources to ensure seamless availability of critical components?
• Supply Ordering: How can AI improve ordering forecasting, ideally combining early warnings and future usage patterns, to strengthen supply resilience while optimizing costs?
• Predictive Maintenance: How can AI predict the likelihood of future hardware failures, suggest spare parts stock levels, and recommend measures to reduce failure risks?
• Supply Autonomy: How can AI support processes like refurbishment, remanufacturing, or repair of equipment and components to reduce dependencies and increase supply resilience?

We are looking for solutions that not only demonstrate technological innovation but also offer practical integration into existing supply chains and procurement processes. A strong focus on sustainable, scalable approaches that consider the full lifecycle of supply chains is essential.

We invite startups, established tech providers, and research initiatives alike to bring forward AI innovations that can significantly enhance the resilience of our supply networks.

​
Innovative Application of AI in Cybersecurity

Threat Detection & Prevention

• How might we use AI to detect zero-day vulnerabilities before they are exploited?
• How might we identify insider threats using behavioral AI models without violating privacy?
• How might we leverage AI to predict and prevent supply chain attacks across interconnected systems?
• How might we use federated learning to detect threats across organizations without sharing sensitive data?
• How might we use AI to detect compromised devices (e.g. spyware, botnet, ransomware, etc.) based on network telemetry (e.g. Netflow/IPFIX, etc.) in our networks?
• How can AI-driven red teams be used to proactively identify vulnerabilities in AI systems, and what safeguards should be in place to prevent the red team itself from creating new risks?

Remediation & Response

• How might we automate incident response using AI-driven playbooks that adapt in real time?
• How might we use reinforcement learning to optimize remediation strategies based on past outcomes?
• How might we enable AI systems to self-heal compromised environments without human intervention?

Advisory Systems & Decision Support

• How might we develop AI-powered advisors to assist CISOs in prioritizing security investments?
• How might we use AI to simulate the impact of security decisions before they are implemented?
• How might we build AI agents that collaborate with human analysts to improve SOC efficiency?

Deception & Offensive Security

• How might we use generative AI to create realistic honeypots that attract and trap attackers?
• How might we deploy autonomous deception networks that evolve based on attacker behavior?
• How might we use AI to simulate red team attacks and uncover hidden vulnerabilities in AI systems?

Strategic & Organizational Security

• How might we use AI to continuously assess and improve an organization’s security posture?
• How might we integrate AI into governance frameworks to ensure ethical and secure AI deployment?
• How might we use AI to detect and mitigate risks from shadow IT and unauthorized AI tools?

Network Traffic & Security Events Data Generation

• How might we use AI to create realistic benign and attack network data (Netflow, IPFix) for testing purposes?
• How might we use AI to create realistic benign and attack security system data (Firewall, IDS/IPS) for testing purposes?
• How might we leverage AI to keep up with new sophisticated attack’s techniques and generate updated data for novel attacks and tactics.

Securing AI Technology & Applications

Prompt Injection & Model Manipulation

• How might we detect and prevent prompt injection attacks in real-time across AI-driven interfaces?
• How might we design AI systems that are resilient to logic-layer prompt control injection (LPCI) without compromising usability?
• How might we use AI to audit and sanitize user inputs to prevent malicious prompt manipulation?

Data Poisoning & Training Integrity

• How might we identify and mitigate data poisoning attempts during model training and fine-tuning?
• How might we use AI to validate the integrity of training datasets in decentralized environments?
• How might we build models that can self-diagnose and recover from poisoned data inputs?

Model Inversion & Privacy Risks

• How might we prevent attackers from reconstructing training data via model inversion techniques?
• How might we use differential privacy and AI to protect sensitive data embedded in model weights?
• How might we detect unauthorized attempts to extract proprietary model knowledge or intellectual property?

AI Governance & Risk Management

• How might we develop AI agents that monitor other AI systems for compliance and security violations?
• How might we use AI to continuously assess the risk posture of deployed AI models across environments?
• How might we create explainable AI systems that can justify decisions and flag suspicious behavior?

Secure Deployment & Runtime Protection

• How might we secure AI models deployed at the edge against tampering and adversarial inputs?
• How might we use AI to monitor runtime behavior of deployed models and detect anomalies?
• How might we build secure containers or sandboxes for AI agents to operate safely in hostile environments?

Agentic AI Security

• How might we design Agentic AI systems to detect and neutralize malicious prompt injections in real time?
• How might we empower Agentic AI to self-monitor for signs of behavioral drift caused by poisoned data?
• How might we prevent Agentic AI from unintentionally revealing sensitive training data through its outputs?
• How might we train Agentic AI to recognize and reject adversarial inputs without compromising usability?
• How might we make the decision-making processes of Agentic AI transparent enough to detect security anomalies?
• How can Agentic AI help detect and respond against multi-layered and diverse cyberattacks.

Summary: Innovative Application of AI in Cybersecurity

AI-Driven Threat Detection & Prevention

• Explore how AI, including behavioral models and federated learning, can detect zero-day exploits, insider threats, compromised devices, and supply chain attacks—without compromising privacy or data sovereignty.
• Investigate the use of AI-generated network telemetry analysis (e.g., Netflow/IPFIX) to identify malware, botnets, and ransomware.
• Assess the role of AI red teams in proactively identifying vulnerabilities in AI systems, while ensuring they don’t introduce new risks.

Autonomous Remediation & Response

• Develop adaptive, AI-driven incident response playbooks that evolve in real time.
• Apply reinforcement learning to optimize remediation strategies based on historical outcomes.
• Enable self-healing AI systems capable of autonomously restoring compromised environments.

AI for Strategic Decision Support

• Build AI-powered advisory systems to help CISOs prioritize investments and simulate the impact of security decisions.
• Design collaborative AI agents that work alongside human analysts to enhance SOC efficiency.

Deception & Offensive AI Security

• Use generative AI to create realistic honeypots and deception networks that evolve with attacker behavior.
• Simulate AI-driven red team attacks to uncover hidden vulnerabilities in both traditional and AI-based systems.

Organizational & Governance-Level AI Security

• Implement AI to continuously assess and improve security posture, detect shadow IT, and manage unauthorized AI tools.
• Integrate AI into governance frameworks to ensure ethical, secure, and compliant AI deployment.

Synthetic Data Generation for Security Testing

• Leverage AI to generate realistic benign and malicious network/security data (e.g., Netflow, IDS/IPS logs) for testing and training.
• Keep pace with evolving threats by generating synthetic data that reflects novel attack techniques and tactics.

Summary: Securing AI Technology & Applications

Prompt Injection & Input Manipulation Defense

• Develop AI systems that detect and prevent prompt injection attacks (including logic-layer control) in real time, while maintaining usability.
• Use AI to audit, sanitize, and validate user inputs to prevent malicious manipulation across interfaces.

Data Poisoning & Training Integrity

• Build mechanisms to identify and mitigate data poisoning during training and fine-tuning, especially in decentralized environments.
• Enable models to self-diagnose and recover from poisoned inputs, ensuring training data integrity.

Privacy Protection & Model Security

• Prevent model inversion attacks and unauthorized extraction of proprietary knowledge.
• Apply techniques like differential privacy to safeguard sensitive data embedded in model weights.

AI Governance & Risk Monitoring

• Design AI agents that monitor other AI systems for compliance, security violations, and behavioral anomalies.
• Continuously assess the risk posture of deployed models and ensure decisions are explainable and auditable.

Secure Deployment & Runtime Protection

• Secure AI models at the edge against tampering and adversarial inputs, using runtime monitoring and anomaly detection.
• Develop secure containers or sandboxes for AI agents to operate safely in hostile environments.

Agentic AI Security

• Empower Agentic AI to detect and neutralize prompt injections, recognize adversarial inputs, and self-monitor for behavioral drift.
• Ensure Agentic AI maintains transparency in decision-making and can respond to multi-layered cyberattacks without compromising usability or privacy.