Beyond AI Apps: Investing in AI Infrastructure

PrimaryMarkets Male

Artificial intelligence may be capturing attention through applications, assistants and automated tools, but the larger investment story could lie beneath the visible layer. In this episode of Unlocking Liquidity, we explore why investors are increasingly looking beyond AI applications to the infrastructure supporting the digital economy. From sovereign computing facilities and connected physical assets to strategic decision-making platforms and advanced holographic visualisation, AI is creating opportunities across software, hardware, energy and critical infrastructure. We also examine several companies currently available through PrimaryMarkets, including Sovereign AI Infrastructure, KERB, Mindhive, Joolie Holdings and Axiom Holographics, and consider the commercial and investment questions sophisticated investors should be asking as AI moves from experimentation to large-scale deployment. ## Chapter

PrimaryMarkets Male

For much of the recent artificial intelligence investment cycle, attention has centred on the application layer. Investors have been presented with AI assistants, automated content platforms, specialist analytics tools and software designed to transform almost every industry. These applications are often easy to understand because they sit close to the end user and promise visible improvements in productivity, cost or customer experience. --- However, the expansion of AI is creating a much broader investment landscape. Every AI application depends on an underlying network of computing power, data storage, energy, connectivity, physical infrastructure and specialised technology. As adoption moves from experimentation to commercial deployment, the systems supporting AI may become just as important as the applications themselves. This is encouraging investors to look beyond the latest AI interface and consider the infrastructure required to make the digital economy work at scale. In private markets, that opportunity extends from sovereign computing facilities and edge-processing technology to the digital management of physical assets and advanced visualisation systems. For sophisticated investors, the challenge is to distinguish businesses that are simply adding an AI label from those solving genuine infrastructure, operational or commercial problems.

PrimaryMarkets Male

The first phase of generative AI was dominated by software demonstrations. The next phase will be shaped by deployment. Businesses and governments now need to determine where their data is held, how models are trained, what computing capacity is available and whether AI systems can be integrated securely into existing operations. This makes AI increasingly capital intensive. Advanced models require specialised processors, high-density computing systems, storage, cooling, secure facilities and significant electricity supply. Applications operating in real time may also need edge infrastructure that can process data near the location where it is generated, rather than relying entirely on a distant cloud environment. The Australian Government’s National AI Plan recognises this shift. Its objectives include building smart infrastructure, strengthening domestic AI capability and attracting global investment. The Government has also introduced expectations for data centre and AI infrastructure developers covering national security, data sovereignty, energy, water, skills and local capability. These expectations reflect the growing recognition that AI -- infrastructure -- is not simply another category of commercial property. It is becoming part of the nation’s economic and strategic architecture. Australia has several characteristics that could support this sector, including political stability, established legal protections, available land, renewable energy potential and proximity to growing Asia-Pacific markets. Government figures indicate that Australia attracted US$6.7 billion in data-centre capital investment during 2024, the second-highest level globally. This infrastructure may also help retain employment, intellectual property and innovation within Australia. The investment opportunity is therefore moving beyond who builds the most popular AI application. It increasingly includes who owns, operates, secures, powers and connects the infrastructure on which those applications depend. ## Chapter

PrimaryMarkets Male

Cloud computing has allowed businesses to access enormous processing capacity without owning their own infrastructure. Yet reliance on large global cloud providers can create concerns for organisations handling sensitive, regulated or strategically important information. Data sovereignty is not merely about the physical location of a server. It also involves the laws governing the data, the parties that can access it, the security arrangements surrounding it and the degree of operational control maintained by the organisation. These considerations are especially relevant in defence, healthcare, financial services, government and critical infrastructure. Organisations operating in these sectors may require private or isolated computing environments, stronger visibility over data handling and assurance that sensitive workloads remain within an approved jurisdiction. One current opportunity on PrimaryMarkets that reflects this theme is Sovereign AI Infrastructure Inc.. The company builds, owns and operates NVIDIA-accelerated facilities designed to allow enterprises and government organisations to train, test and deploy private AI under local control. Its model combines computing capacity with owned sites and secure, onshore operating environments. --- This illustrates an important change in how some AI businesses may be assessed. Rather than relying entirely on projected software subscriptions or future user growth, infrastructure businesses can combine technology exposure with physical assets, installed computing capacity and contracted enterprise demand. That combination does not remove risk. Computing hardware can become obsolete, facilities require substantial capital and electricity, and expansion must be matched with sufficient customer utilisation. Nevertheless, the model demonstrates how AI exposure can extend beyond application developers to the owners and operators of the underlying infrastructure.

PrimaryMarkets Male

AI infrastructure is not confined to data centres. It also includes the technologies that allow buildings, vehicles, equipment and other physical assets to generate, communicate and act on information. Many valuable assets still operate with fragmented or outdated systems. Car parks, commercial buildings, ports, airports and industrial facilities may contain physical equipment that performs its basic function but produces limited usable data. Replacing all that equipment can be expensive and disruptive. An alternative is to install a digital layer that connects established infrastructure to cloud platforms, analytics and automated decision-making. KERB, another company currently raising through PrimaryMarkets, provides an example of this approach. The company describes its platform as a parking operating system for buildings and infrastructure. It is designed to connect existing parking equipment with a cloud-based platform that provides information on occupancy, revenue, access, vehicle activity and operational performance. KERB’s hardware-agnostic model is intended to allow property owners, airports, ports, campuses and car park operators to modernise existing assets without replacing their core infrastructure. The company is also developing edge-AI vehicle fingerprinting technology, demonstrating how processing and computer vision can be brought closer to the physical point of activity. --- The broader investment theme is larger than parking. Across the economy, physical assets are acquiring a digital layer. Buildings are becoming more responsive, transport infrastructure more measurable and industrial equipment more connected. Businesses able to integrate with legacy assets may gain an advantage because customers can adopt new capabilities incrementally rather than funding a complete replacement. For investors, the critical questions concern compatibility, implementation costs and measurable customer benefits. A platform may have impressive technical capabilities, but its commercial value depends on whether it can increase asset utilisation, reduce costs, improve security or unlock new revenue.

PrimaryMarkets Male

Looking beyond AI applications does not mean applications themselves are no longer attractive. It means investors may need to examine them with greater discipline. The rapid availability of foundation models has made it easier to develop AI- enabled software. This can reduce development time, but it can also lower barriers to entry. A new application may attract early attention without possessing proprietary data, meaningful customer relationships or a durable competitive advantage. The more compelling application businesses are likely to be those addressing expensive, repetitive and poorly served workflows. They must demonstrate that AI is performing valuable work rather than merely adding a conversational interface to an existing service. <u>Mindhive</u> provides another example at the application layer. The platform helps organisations run large-scale strategic planning, problem-solving and decision-making processes using both human experts and AI. Its approach illustrates how AI can support complex organisational challenges by combining machine-assisted analysis with structured input from experienced contributors. -Joolie Holdings provides an example at the application layer. Joolie is developing an AI property manager for self-managing landlords, automating activities including leasing, tenant screening, rent collection, maintenance coordination, inspections and compliance tracking. Its relevance to the infrastructure theme comes from its connection to a substantial physical asset class. Residential property management involves recurring administrative processes, regulatory requirements and communication between landlords, tenants and service providers. An effective AI platform could become part of the operational infrastructure supporting these assets. The investment question is not simply whether the technology works. It is whether landlords will pay for it, whether the cost of acquiring customers is sustainable and whether the platform can maintain accuracy across different tenancy laws and market conditions. Investors must also consider the extent to which accumulated property and compliance data could improve retention or create defensibility over time.

PrimaryMarkets Male

The digital economy will also require better ways to interpret and interact with complex information. AI can generate models, simulations and enormous volumes of data, but value is only created when people can understand and use the results. Advanced visualisation may therefore become an important part of the infrastructure surrounding AI, engineering, defence, education, design and entertainment. Three-dimensional interfaces can allow users to explore information that is difficult to communicate through conventional screens. Axiom Holographics is an Australian technology company developing holographic devices and rooms that create three-dimensional objects from light. Its clients include organisations across aerospace, defence, energy, museums and hospitality. --- Axiom’s inclusion within the broader digital infrastructure theme highlights the relationship between computing and human interaction. As digital models become more sophisticated, organisations may require improved tools for training, planning, simulation, product presentation and collaborative decision-making. Investors considering advanced hardware businesses must nevertheless account for manufacturing risk, installation capacity, supply chains and the capital required to expand internationally. A strong order book is valuable, but the ability to manufacture, install and support products efficiently ultimately determines whether demand converts into sustainable earnings.

PrimaryMarkets Male

Infrastructure-oriented technology businesses require a different analytical approach from conventional software companies. Investors should consider the durability of the underlying demand, the capital required to meet it and the company’s position within a rapidly changing technology stack. Ownership of physical assets can provide strategic value, but it can also increase funding requirements. Recurring software revenue can improve visibility, but only where customer retention and implementation economics are attractive. Proprietary technology can create a competitive advantage, although that advantage must be assessed against the pace of innovation and the possibility of larger competitors entering the market. Energy access is becoming particularly important. The Australian Government expects new data centres and AI infrastructure to contribute to additional renewable supply, pay their appropriate grid-connection costs and use demand flexibility to avoid placing unnecessary pressure on households and other businesses. Energy, water and community acceptance are therefore becoming investment considerations rather than peripheral sustainability matters. Investors should also separate market potential from company execution. A large addressable market does not guarantee that a particular business will capture it. Management capability, customer concentration, funding structure, --- valuation, intellectual property, regulatory exposure and exit pathways remain essential elements of due diligence.

PrimaryMarkets Male

The most visible AI businesses are not necessarily the only beneficiaries of the technology’s growth. As AI becomes embedded across commerce, government and industry, demand is expanding throughout the supporting ecosystem. Computing facilities, sovereign data environments, connected physical assets, edge processing, workflow automation and advanced visualisation all form part of this emerging infrastructure. Some opportunities will resemble traditional technology companies, while others will combine software, hardware, property and recurring services. For sophisticated investors, this creates a wider field of potential exposure but also requires more careful comparison. The central question is no longer simply whether a company uses AI. It is whether the business owns or controls something necessary, solves a commercially significant problem and possesses a credible model for converting adoption into sustainable value. PrimaryMarkets provides wholesale, sophisticated and institutional investors with access to a range of private capital-raising opportunities across AI infrastructure, technology and digitally enabled industries. Investors can review the current opportunities and supporting information through the PrimaryMarkets capital raising platform.

PrimaryMarkets Male

And that brings us to the end of this episode of Unlocking Liquidity. Thanks for spending your time with us, we hope today’s conversation gave you a fresh perspective on private markets and how liquidity is evolving. If you enjoyed the episode, please follow or subscribe wherever you listen, and feel free to share it with someone who’d get value from it. For more insights, opportunities and episodes, visit PrimaryMarkets.com. Until next time, thanks for listening, and we’ll see you in the next conversation.