Global X AI and Innovative Technology Active ETF (3006) Outlook 2025

It has been two years since the release of GPT3.5 to the public in November 2022. Since then, the launch triggered an arms race to build the most advanced LLMs (large language models). The competitive landscape intensified quickly. Tech giants like Google, Microsoft, and Meta unveiled their own advanced models—PaLM, Gemini, and LLaMA—while startups such as Anthropic and Mistral AI gained traction with specialized, safety-focused systems. Investment poured into both training and inference capabilities, reflecting the astronomical costs of building and deploying LLMs. Model performance continues to increase with dropping costs. This democratization of AI sparked a wave of adoption across sectors such as customer support, content creation, and software development. The AI LLM sector has become a hub for investment, innovation, and competition. As companies push the boundaries of what these models can achieve, the broader implications—on jobs, governance, and global power dynamics—continue to unfold, marking this as one of the most transformative periods in technology history. Looking into 2025, the key focus of the sector centers around two points. First, the constraint of scaling in front-tier models and its implications for AI infrastructure spending. Second, adoption and monetisation of LLMs.

Scaling of Large Language Model and its Implications to AI Infrastructure

Research indicates that larger models typically exhibit improved performance, as evidenced by the “scaling laws” identified by Kaplan et al. (2020), which suggest that increasing model size often leads to better accuracy and functionality in diverse applications.

Scaling LLMs involves not just increasing the number of parameters, but also optimizing training processes and leveraging advanced hardware architectures. This necessitates robust AI infrastructure encompassing high-performance computing resources, efficient energy consumption, and scalable storage solutions. Scaling laws suggest that next-generation front-tier models (ChatGPT 5 equivalent) require $3B in GPU investment. The generations afterwards ( ChatGPT 6/7) could cost 10x-20x, or $25B+-$50B+ each. (MS 2024) The demand for AI training and inference semiconductors from companies like Nvidia and other ASIC providers like Broadcom and Marvell will therefore continue to increase if the industry scales the models under the current trajectory.

Data availability is arguably the most important constraint in the scaling of LLMs as we only have one internet. Training large AI models requires correspondingly large datasets. The indexed web contains about 500 trillion words of unique text and is projected to increase by 50% by 2030. (Epoch AI 2024) Multimodal learning from image, video and audio data will likely moderately contribute to scaling, plausibly tripling the data available for training. Synthetic data generation from AI models could increase data availability substantially.

Adoption of LLMs – Impact to the Software Industry

The integration of Generative AI (Gen AI) into the software landscape is reshaping how technology companies deliver value across various layers. By examining the Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and Software as a Service (SaaS) layers, we can better understand the evolving strategies that companies, especially giants like Microsoft and Amazon, are deploying to capitalize on Gen AI. Each layer presents distinct opportunities and challenges, influencing how organizations can generate revenue and drive innovation.

IaaS serves as the foundational layer where cloud computing resources—such as computing power, storage, and networking—are made available to users. At this stage, IaaS has been the focus of AI investment so far, where the industry has invested sizable hardware to support LLM training and inference workload.

Platform-as-a-Service (PaaS) is poised to capture increasing value as the focus shifts to the platform layer, which will serve as the foundation for enabling practical, scalable generative AI use cases at the application layer. The growing complexity of IT environments within enterprises is expected to remain a significant bottleneck in the adoption of AI-native applications. This creates a compelling opportunity for both established players and emerging vendors to deliver solutions that simplify and streamline the deployment and integration of generative AI technologies.

SaaS represents the final layer where end-users directly interact with AI-enhanced applications. The potential for Gen AI to transform SaaS offerings is considerable, as applications can become more responsive, personalized, and capable of automating complex tasks. For instance, companies like Salesforce and Adobe are already embedding Gen AI into their applications, enabling features such as predictive analytics and automated content generation.

Global X AI and Innovative Technology Active ETF (3006) Strategy

The Global X AI and Innovative Technology Active ETF Fund is strategically positioned to capture the full value chain of AI development, from foundational infrastructure to transformative applications. Our investment approach is rooted in identifying opportunities across the core verticals globally: from AI Infrastructure to application layers. By maintaining a diversified yet focused portfolio, we aim to deliver long-term capital growth while managing risks associated with this rapidly evolving sector.

Our fund’s core holdings currently center around AI infrastructure providers or supply chains like Microsoft, Alphabet, Meta, TSMC, and Nvidia. These companies represent the backbone of AI development, providing essential hardware, software, and services that enable AI research and deployment. The competitive landscape in this area is more predictable at this stage given the significant CAPEX required and technology barriers in semiconductors. These barriers create a high moat for existing leaders, allowing them to capitalize on their technological advancements and scale.

In addition to AI infrastructure, the fund also gains exposure to AI robotics through Tesla. We view Tesla as a global leader in AI robotics with deep expertise in full-stack autonomous driving and humanoid robotic system development. The company’s ongoing innovations in AI and robotics align with our strategy to invest in transformative technologies that have the potential to disrupt traditional industries and create new market opportunities.

Over time, we look to increase our exposure to the platform and application layers as the industry builds out the ecosystem and we gain more visibility on the Total Addressable Market (TAM) and competitive dynamics in AI downstream sectors.

The Global X AI and Innovative Technology Active ETF Fund is committed to being at the forefront of AI investment, leveraging our expertise to identify and capitalize on opportunities across the AI value chain. By focusing on both established leaders and emerging innovators, we aim to provide our investors exposure to one of the most dynamic and impactful sectors of the global economy. As the AI landscape continues to evolve, we remain dedicated to adapting our strategy to ensure that our investors benefit from the full spectrum of AI-driven growth and innovation.