Megaport's AI Compute Pivot Splits Fund Managers After a 208% Re-Rating

A network operator turned AI infrastructure play has some fund managers still buying and others already moving on.

6 Oct 2026 · 5 min read

Megaport spent the June quarter turning itself from a network connectivity provider into one of the ASX's most talked about AI infrastructure plays, re-rating roughly 208 percent in three months before giving back some of those gains in a broader AI sell-off in July. The pivot has split fund managers into three camps: those who think the market still hasn't fully priced the opportunity, those who backed the raise but now think the easy money has been made, and those avoiding the stock altogether on execution risk.

From connectivity to compute

The catalyst was Megaport's 2025 acquisition of Latitude.sh, which several managers said was finally being monetised. Centennial Asset Management noted the company announced new contracts worth $458.9 million alongside an $827.3 million capital raise, with new shares issued at $14.30 against a June 30 share price of $21.58. Paradice Investment Management, which does not hold the stock, put pro forma compute annualised recurring revenue at roughly A$385 million and total contract value from the new wins at about A$748 million. Perennial Partners linked part of the move to roughly A$200 million of new annual recurring revenue from GPU compute contracts signed with non-hyperscaler enterprise customers, while Blackwattle Investment Partners pointed to four new AI infrastructure contracts in June and the launch of an on-demand GPU pool as the immediate drivers.

Yarra Capital Management, which holds an overweight position, said Megaport's global compute division won a number of large CPU and GPU clients during the period and that the balance sheet remains in a net cash position, giving it capacity to fund further customer wins across both compute and its software-defined networking business.

The bull case: an underappreciated AI infrastructure play

Quest Asset Partners, which added the stock in April during an AI-driven tech sell-off, argued the combination of Megaport's existing network footprint and its new compute offering is genuinely differentiated, and that the recent capital raise gives it firepower to deploy into compute at high returns. The fund called it simply the best AI exposure on the ASX. Wilson Asset Management rated the stock a buy in July, arguing chief executive Michael Reid is among the highest quality technology chief executives on the ASX and that Megaport trades on around half the valuation of its US peer DigitalOcean, with industry feedback suggesting the company is in the market for more than a billion dollars of debt to fund further growth, a signal Wilson read as implying substantial earnings upgrades ahead.

Firetrail Investments made a similar case in more granular terms, pointing to Megaport's presence in around 1,100 data centres across 31 countries and forecasting that AI inference demand could grow more than 400 percent by 2030. When Megaport was later caught up in a broader, indiscriminate sell-off of AI-exposed names over concerns about open-source competition and capital expenditure sustainability, Firetrail said it saw no deterioration in the underlying business, pointing instead to AMD lifting its own data-centre CPU market forecast by 80 percent as the standout positive signal in the noise.

With a presence in 1,100 data centres across 31 countries and AI inference demand forecast to grow over 400% by 2030, the fund expects further debt-funded contract wins ahead.

— Firetrail Investments, Monthly Report, July 2026

Not everyone thinks the rally has further to run

Pendal Group summed up the more cautious turn in its July commentary, calling itself a great admirer of the business and its underappreciated 1,000-site global data centre network, but moving to a hold rating after the stock ran hard north of $20 a share, on the view that better value could be found elsewhere. Paradice Investment Management, watching from the sidelines, respected the quality of Megaport's network assets but said it could not justify the valuation ahead of the news flow, and argued the price after the re-rating embeds execution assumptions that leave little margin for error.

We believe the current price embeds execution assumptions that leave little margin for error.

— Paradice Investment Management, Quarterly Report, June 2026

Lennox Capital Partners went further, deliberately avoiding Megaport despite the rally and citing execution risk in scaling the newly acquired Bare Metal capability into a much larger contracted commitment, preferring its unlisted holding in Firmus as its chosen route to AI compute growth instead. Elston Asset Management, which fully participated in the capital raise and called Megaport a standout during the quarter, nonetheless trimmed its position afterward to manage risk and lock in gains, while Quest Asset Partners' monthly commentary noted it had similarly trimmed its holding once a guidance upgrade had already lifted the shares.

July brought a reminder of how quickly sentiment can turn. Lennox counted avoiding speculative AI-adjacent names, Megaport among them, as a positive contributor to relative performance during a month when the stock fell around 16 percent in a broader unwind of momentum and AI-linked names. Blackwattle described the pullback as a normal consolidation after a period of rapid appreciation rather than a change in the underlying story, a view echoed by Firetrail's read that fundamentals remained intact.

What happens next looks likely to hinge on execution rather than further re-rating. Fund managers say they are watching whether Megaport can convert its expanding pipeline of compute contracts into recurring revenue without cost or delivery slippage, how the flagged debt raising to fund hardware is structured and priced, and whether sentiment toward AI infrastructure names more broadly stabilises after July's sell-off. For skeptics like Paradice and Lennox, the bar is proving the newly acquired compute business can scale as promised. For bulls like Quest, Wilson and Firetrail, the argument is that the market is still under-crediting a network operator that has found a much bigger addressable market than the one it started with.