By Ron Shamgar

Megaport is a company that provides the plumbing for the internet and cloud computing, but made simple and easy to use. Over the past 14 years it has built a large, software-controlled network that connects businesses to cloud services in more than 1,100 data centres across 30+ countries. More recently, Megaport bought a company called Latitude that lets customers quickly spin up computing power (CPUs and GPUs) and storage in those same locations through an online portal.

What the company offers today

Fast, on-demand access to compute (CPUs and GPUs), storage, and high-speed networking delivered through software. Customers can provision servers in seconds.

A global footprint of data centres and existing contracts, so Megaport can place equipment in many locations quickly instead of building new giant data centres.

An integrated product: network plus compute plus storage all automated and managed from one platform.

Why AI matters to Megaport

New AI tools (large language models, coding assistants, and “agents” that automate tasks) are creating huge demand for two things: fast computers called GPUs, and lots of supporting CPU and network capacity.

There are two kinds of AI work. Training big models requires massive, power-hungry data centres. Inference, which is running models to serve users, can be spread across many locations. Inference suits Megaport’s distributed setup.

Many companies cannot get GPUs quickly from big cloud providers because supply is tight. That creates strong demand for the kind of on-demand GPU capacity Megaport can provide.

Recent wins and what they mean

Megaport has announced several large contracts totalling hundreds of millions of dollars and is raising capital to deliver those contracts and build an on-demand GPU pool.

The company already runs thousands of CPUs and GPUs via Latitude. The planned GPU pool will let customers rent GPU capacity by the hour or month, giving quick access without long waits.

These contracts and the GPU pool act like a “shock absorber”. Customers can scale quickly when they need extra capacity while Megaport procures and deploys longer-term capacity.

The opportunity ahead

Short term: meet urgent customer demand for GPUs and compute that big cloud providers cannot satisfy fast enough.

Medium term: attract new customers using on-demand access, then convert some to longer contracts while cross-selling networking and storage.

Long term: support enterprises running private or regulated AI systems by offering secure, geographically distributed compute plus the network to connect it.

In short, Megaport has combined a huge global network with easy-to-use compute and storage tools. With strong AI-driven demand and recent large contracts, it is well positioned to help companies deploy and scale AI applications quickly.

The company is on track to annualise $660m of ARR and is growing very fast. At a valuation of $3B we think the upside is significant. Megaport offers one of the better AI exposures on the ASX. With reported rumours of the upcoming listings of Firmus and others, we prefer an established player, with a track record of execution, profitability and a reasonable valuation.

The risks investors should think about

A thesis like this would not be honest without acknowledging where it could go wrong.

The first risk is the capital raise itself. The $827m entitlement offer is priced at $14.30, a 13.9% discount to the last close. Shareholders who do not participate will be diluted, and existing investors are absorbing a meaningful equity issuance to fund a deployment that has not yet been earned out on the income statement. The raise is fully underwritten, which removes execution risk on the funding side, but it does not remove the dilution.

The second risk is utilisation. The $350m GPU Pool is the speculative part of the strategy. Unlike the four contracted deals, the GPU Pool is capacity that Megaport is deploying before customer demand is locked in. If on-demand demand falls short of expectations, the company carries the utilisation risk directly. That is a different kind of exposure to a contracted revenue stream and the market will watch utilisation metrics closely over the next 12 to 18 months.

The third risk is capex intensity. Building out high performance NVIDIA GPU infrastructure is not cheap, and Megaport has committed approximately $369.5m of capex to service the new contracts on top of the GPU Pool investment. The historical Megaport investment case has been about an asset-light, software-defined network platform. The new model is more capital intensive, and that has implications for return on invested capital that investors should be comfortable with before assuming the same multiple applies to the new business.

The fourth risk is competition. Hyperscalers like AWS, Azure and Google Cloud are not standing still. If they resolve their GPU supply constraints faster than expected, or if they aggressively compete on price and distribution for inference workloads, the window Megaport is moving into could narrow. Specialised AI infrastructure providers like CoreWeave, Lambda Labs and others are also targeting the same enterprise customers.

The fifth risk is the broader AI cycle. The bull case for Megaport assumes that AI inference demand continues to scale rapidly and that enterprises continue to need distributed, low latency infrastructure to deploy AI in production. If the AI capex cycle slows, or if inference workloads consolidate into a smaller number of hyperscale locations rather than spreading geographically, the demand profile changes.

None of these risks invalidate the thesis. They are the risks that need to be priced into the position size and the time horizon.

TAMIM Takeaway

The market has been searching for the right way to express AI exposure on the ASX for the last 18 months. Most of the obvious candidates are either expensive (NXT), early stage and speculative (the upcoming Firmus IPO and similar names), or only loosely connected to the AI capex cycle.

Megaport offers something different. An established business with a track record of execution. A profitable core network division that is now growing 25 per cent year on year. A clear and credible path into AI infrastructure through Latitude and the GPU Pool. Pro forma group ARR tracking toward $660m. A capital raise that funds real, contracted customer demand rather than a hopeful expansion into a new market. And a valuation that, while no longer cheap, has not yet priced in the full optionality of the global inference cloud opportunity.

For long term investors looking for AI exposure on the ASX, the practical question is not whether Megaport is the cheapest way to play the theme. It is whether it is the best risk-adjusted way. We think the combination of an existing network business, a credible AI strategy, an experienced management team, and a reasonable valuation makes it one of the more compelling AI exposures available on the local market today. The risks are real and the execution bar is high, but the setup is one of the better ones we have seen.

______________________________________________________________________________________________

Disclaimer: Megaport (ASX: MP1) is held in TAMIM portfolios as at the date of article publication. Holdings can change substantially at any time.