Megaport at $338M ARR: What the 49% Growth Line Hasn’t Priced

Annual recurring revenue of $338M, up 49% year on year, frames everything else at Megaport right now. The stock has run 133% in three months and added another 15.18% in the latest session, closing at $18.48. That kind of move usually invites the reflex to fade it. But ARR accelerated while the price moved, and a re-rating that tracks an accelerating top line is a different animal from one driven by multiple expansion alone. The half-year numbers gave the market a reason, and I’d argue the reason is still underpriced.

Start with retention, because in this category expansion economics decide whether growth compounds or leaks. Net revenue retention is 111%, meaning the existing logo cohort spent 11% more than it did a year earlier before a single new customer is counted. That is the floor under the 49% ARR figure. A usage-based connectivity platform with NRR above 110% does not need to win the same customer twice — it lands an enterprise on a virtual cross-connect, then upsells additional ports, cloud routers, and edge nodes as that customer’s multi-cloud footprint grows. Revenue rose 26% to $134.9M over the half, and the gap between that and the 49% ARR growth tells you the recognized number is trailing the contracted number, not leading it.

Then the contract wins did something the retention line alone could not. In mid-May the company secured three US tech contracts worth $254M in total contract value, adding roughly $90.6M in ARR in a single announcement — more than a quarter of the entire installed ARR base, booked at once. Early June added four AI infrastructure contracts and an $827M capital raise aimed at an inference cloud build-out. The capital raise is the part worth sitting with: a company does not raise that scale to defend a position, it raises to fund deployment ahead of demand it has already signed. The full-year Megaport Network revenue guidance was revised up, with the lower end lifted to $264-270M. Guidance moving up while ARR accelerates is the combination that justifies paying a higher EV/ARR than the stock carried at $7.93.

The thing the market tends to miss in names like this is the order of operations. A usage-priced network platform looks expensive on any trailing multiple right up until the contracted ARR converts into recognized revenue, at which point the multiple compresses on its own without the stock moving. The setup that keeps recurring across software infrastructure is identical here: an installed base with high gross retention and a magic number that says sales are landing efficiently, sitting on contracted bookings the income statement has not caught up to. Investors who waited for the recognized revenue to validate the ARR line have historically paid up for the same business after the cheap window closed. The window here is the lag between $338M of ARR and the $134.9M half-year revenue run-rate.

Macro matters too: the June US executive order on advanced AI innovation and security is pulling forward private-sector infrastructure deployment by stripping regulatory friction off frontier-model and data-center build-outs, which is precisely the connectivity Megaport’s cross-connects and edge nodes sit underneath. In Europe, the Commission’s Cloud and AI Development Act and the broader sovereignty push are forcing enterprises toward neutral, software-defined networking that can satisfy data-residency rules across multiple clouds — a vendor-neutral platform is the structural answer to a sovereign-cloud mandate. AI inference workloads need low-latency interconnection between compute, storage, and cloud regions, and that interconnection is metered and recurring. The macro is not a tailwind in the abstract; it is the specific reason the May and June contracts landed when they did.

Local structure adds one more edge. Founded and listed in Australia, Megaport built a neutral platform across more than 1,100 enabled data centers before the US hyperscalers turned interconnection into a competitive battleground, which is why the recent AI contracts went to a network-as-a-service provider rather than a single cloud’s native fabric.

None of this holds if the conversion stalls. The clearest risk is the one buried in the retention figure: 111% NRR is healthy, but it has room to fall, and if the existing cohort’s expansion slows, the 49% ARR growth decelerates toward the 26% revenue print rather than the revenue catching up to ARR. The $827M raise also dilutes — funding an inference cloud is capital-intensive in a way the asset-light cross-connect business historically was not, and if the compute build-out runs at a lower payback period than the core network, the blended return on that capital drops. There is concentration risk too: a single $90.6M ARR tranche from three customers means the next contract cycle has to keep delivering at that magnitude to sustain the growth rate, and US tech capex is the variable that decides it. If ARR growth decelerates below 30% year on year while NRR slips under 105% over the next two reported halves, the re-rating loses its fundamental anchor and this case breaks down.

At $18.48 the price-derived multiples now sit well above where the trailing fundamentals would put them, because the reported revenue base beneath them has not yet moved with the price. That gap is the entire argument. The bookings are signed, the guidance is raised, and the recognized revenue is the slowest variable in the system. I’d rather own the lag than wait for it to close.

Figures reflect the most recent available data and may differ slightly from live market prices. · © Mathstock