Celadon Partners Secures First Close on Data Centre Fund

Celadon Partners is focusing on data centre value chain opportunities, having reached a first close of approximately USD 200m on its debut blind pool fund, said CEO Donald Tang. A trans-Pacific investor headquartered in the US and with offices in Asia, the firm focuses on mid-market buyouts or minority situations where it can drive operational transformation. Since its founding in 2018, Celadon has deployed capital from its balance sheet and separately managed accounts across artificial intelligence (AI) hardware, consumer, healthcare, and business and financial services. “We find fundamental trends that are interesting, but we don’t chase the most direct expressions of those trends. We look one or two degrees removed, at areas that are adjacent or auxiliary,” said Tang, who previously spent 13 years as Asia CEO of DE Shaw. “Often, that means going upstream and looking at suppliers. These businesses benefit from the fundamentals, but they are often overlooked and undervalued versus the obvious candidates that have everything already priced in.” Celadon decided to raise a blind pool fund – the target is USD 500m, the hard cap is USD 650m – because pipeline deals are increasing in size. This is partly driven by opportunities with large-scale data centre operators. The firm dips into Asia-centric supply chains, for example, helping US partners add new manufacturers to procurement programmes in highly concentrated product areas. “There are companies in markets like Japan that are critical to the AI [artificial intelligence] infrastructure and hardware boom, but they don’t market their story in the most effective way. They have nice businesses, but they are relatively conservative, many degrees removed from the end hyperscaler customer, and trading at single-digit EBITDA multiples,” said Tang. “Hyperscalers have huge supply chain teams, they realise where the issues are, but they don’t have good partners to help solve these issues.” GPUs account for much of the value in AI hardware, but Celadon avoids semiconductors because of geopolitical sensitivities. Areas of interest range from optical networking equipment – a replacement for copper wire, which becomes impractical at distances above two metres – to liquid cooling and power. It is not simply a case of connecting mid-market suppliers with sources of demand. Hyperscalers may seek third-party support when operating in challenging geographies or when looking to build relationships with companies that, while in possession of attractive technology, have yet to demonstrate the ability to deliver at scale and on time. “Data centres are well-oiled machines. It might be a USD 30bn-USD 40bn investment, and if some component doesn’t show up, you’ve got USD 10bn-USD 15bn of GPUs sitting there, depreciating. A new generation of GPUs comes out every 18-24 months, so losing one quarter is imaginable,” said Tang. Half of the approximately 10 investments Celadon made prior to launching the new fund relate to the data centre value chain. Tang is confident that opportunities will remain rich and entry valuations relatively attractive for at least for the next two years. Should the market shift, the firm will revert to its roots, targeting traditional industries where it sees scope for transformation. One deal in this vein was a footwear manufacturer that re-focused on fast turnaround, small batch production runs. This enabled it to serve limited-edition luxury sneaker brands – such as Yeezy, which was launched by US rap artist Ye in conjunction with Adidas – which commanded retail-style operating margins as opposed to the mid-teen margins associated with wholesale. “We invested at a low single-digit EBITDA multiple and grew earnings by 6x or 7x. It’s about selling more efficiently at much higher margins. With limited-edition sneakers, you don’t need large physical retail spaces and inventory levels, so the brand can sell direct and capture the full retail margin without taking on much risk,” said Tang. “Transforming traditional businesses with new business models is just lower risk, and yet in the upside case, the returns aren’t unlike successful tech investments.” Celadon has grown the value of those initial investments more than 70x, albeit by writing relatively small cheques and being creative on structuring. In the footwear deal, for example, customer contracts were restructured as long-term take-or-pay, enabling trade finance via commercial banks. It received a 40% equity stake, but the investment was 25% equity, 75% trade finance, lowering the entry multiple. Rather than just ask for a put option as downside protection, Celadon looks for ways to lock in disproportionate upside as well. It might invest a certain amount upfront and negotiate the right to double the size of its commitment a year later at the same price if it has been successful in helping the company reach certain commercial milestones. “You must present it in a way that makes sense. Say a company is hitting its growth numbers and will need more capital, but it doesn’t want to spend a lot of time raising the money. We will commit to it with valuations locked in upfront. But if the company doesn’t hit its numbers, we have the right to walk,” Tang explained. These numbers might be harder to replicate at scale, but Celadon believes the cross-border opportunity set has expanded for those able to navigate fragmenting supply chains, trade policy, and geopolitics. Staying close to the customer – whether a data centre operator or an original equipment manufacturer (OEM) – and understanding their evolving needs is key. “We don’t invest in the hope that someone will work with us,” Tang added. “In most cases, we have already found the end demand and we can say, ‘This is the opportunity, this is how much business we will be doing together.’”