- Vertically integrated builder-operator with its own IP of around 600 patents tied to data centre designs and systems
- Headcount rising, Malaysia at the centre of that growth with BDC expecting to almost double country workforce in 2026
Outside of building semiconductor fabrication plants, data centres (DC), or AI factories, as Nvidia’s founder Jensen Hwang, has taken to calling them now, is the hottest business in the AI era we are entering. How hot? An April 2025 report from McKinsey threw up the mindboggling US$6.7 trillion (RM26.4 trillion) in estimated capex needed for data centres by 2030.
That puts data centre builders (offering build-to-suit data centres for customers) and operators (where they offer colocation solutions to multiple clients), like Bridge Data Centres (BDC) in a sweet spot. Founded in 2017 in Singapore by data centre veterans Kris Kumar and Michael Foust, BDC is a Southeast Asia–based hyperscale and wholesale data centre provider backed by Bain Capital, a US private investment firm with a global presence.
What is interesting here is that, within the same year that Bain Capital invested in them, the global investment firm with US$215 billion (RM860 billion) in assets under management, made an outright acquisition of BDC for an undisclosed amount. It is rare for founders to sell their companies within one year.
According to an executive of a Malaysian DC operator, for most DC operators who focus on winning business from the hyperscalers, the business is straightforward: secure land and power, build the DC, hand the facility to vendors, then collect rent. But BDC has taken a very different route in an effort to differentiate itself in the market. It adopted a strategy of becoming a vertically integrated builder-operator with its own Intellectual Property – with around 600 patents tied to its data centre designs and systems – and engineering expertise, wrapped around every megawatt it delivers.
This strategy shaped everything from BDC’s supply chain, which is concentrated in China, to its talent strategy, to how it rides the AI wave that is driving the current phase of demand.
A builder, not a landlord
Eric Fan (pic), BDC’s CEO, is clear that he does not see BDC as a real estate play. He points to its pure R&D team of around 20 specialists who sit apart from its regular engineers and project delivery staff. Their job is to work on “frontier solutions” and co-develop new designs with suppliers, then protect that know-how through an aggressive patent strategy.
“BDC today has around 600 patents related to data centre designs and systems. Many actually arise from joint work with long-term suppliers, but BDC is careful about how ownership is structured so its solutions cannot be copied wholesale by rivals. “We’d be able to protect our knowledge and our know-how in those designs, so it’s not going to be easily replicated by our peers,” Fan explains.
“If you position yourself as a pure real estate developer, you don’t need that kind of in-house know-how and technology,” Fan notes. “It’s more about you getting land and power, and you can leave it in the hands of your suppliers. But we do believe differently.” That belief is why BDC is prepared to carry the higher fixed costs of in-house R&D and engineering.
When asked if he can put an estimated ROI value on the patents, he says he cannot. But that is not the point. According to him, the value is in the cumulative edge BDC gets from optimising its designs over time: standardised modules, better energy efficiency, lower lifecycle cost, and the ability to roll out campuses at hyperscaler speed. “Patents are reviewed regularly, with an inflow and outflow process to decide which to maintain and which to drop,” he said. This approach helps keep the IP portfolio relevant.
The key to its vertically integrated strategy is a tightly managed supply chain centred on China. Most of BDC’s modular equipment integration and modularisation work is done there, leveraging the country’s deep manufacturing base and the ability to move from components to pre-assembled modules seamlessly.
While Fan acknowledges the geopolitical and concentration risks, he says BDC is potentially looking at shifting parts of the work into Southeast Asia.
While China’s expertise in manufacturing is obvious, Fan says, “We do see the potential for de-modularization where some of the steps of these design and manufacturing could be done in Southeast Asia countries.” But it will be selective, and he does not foresee a full duplicate manufacturing plant in the region, which in his view would not make economic sense.
Building a regional talent engine
BDC’s integrated model only works if you can attract and grow enough engineers. Out of a group headcount approaching 1,000, BDC has close to 100 engineers across various capabilities, from electrical and mechanical to design and project execution.
Crucially, operations are not outsourced. Unlike operators who hand over day-to-day maintenance to third parties, BDC insists on running its own facilities. In a typical data centre, Fan estimates that “easily” 70% of the onsite headcount are knowledge workers with electrical or mechanical backgrounds, able to troubleshoot and maintain major systems. Security and facilities staff make up the balance.
In Malaysia, which already hosts BDC design teams and a central office in Kuala Lumpur, that translates into a sizable cluster of design engineers and project managers who support the entire region, not just local builds. The design function is distributed across Singapore, Malaysia, and China, operating as one virtual platform that serves all markets. Location is about access to talent, not carving the region into silos.
This is also where BDC’s education partnerships come in. The company recently signed an MoU with the Singapore Institute of Management (SIM) with a DC training module already introduced. While BDC does not have a DC in Singapore, students from SIM will do their internships in BDC’s campuses in Malaysia and Thailand, gaining first-hand exposure to large-scale operations.
Headcount is rising quickly, and Malaysia is at the centre of that growth says Fan with BDC expecting to almost double its Malaysian workforce in 2026, driven by rapid expansion in Johor and the Klang Valley as new DC campuses come online this year. New projects in Thailand and additional technology design roles in Singapore will add to the regional spread, but “the sheer numbers”, as Fan puts it, will still be in Malaysia.
AI demand is surging, but hard to put an estimate on its boost to the business
Any DC narrative today comes with an AI link, and BDC is no exception. The company already has over 300 megawatts of live capacity in Malaysia alone, making it the country’s largest operator, and is building an even larger pipeline to meet future demand. “A very strong pipeline,” as Fan puts it. AI is clearly one of the drivers.
Yet when asked what percentage of BDC’s business comes from AI-related workloads, Fan cautions that it cannot be cleanly measured. According to him, the old distinction between AI training clusters and “regular” workloads is breaking down.
Previously, you could point to a cluster and declare: this is an AI training job. Now, with AI inference embedded into more and more applications, those AI cycles are spread across all kinds of workloads. “It’s harder and harder to define that this is non-AI and this is AI,” he says. “A lot of applications are now infused with AI adoption.”
What he is willing to say is that AI-linked demand is one of the fastest-growing segments in BDC’s portfolio, even if the baseline is still small relative to its total installed base. That growth is clearest in the hyperscaler segment, where cloud and AI platform providers are racing to expand capacity in key hubs like Johor to serve regional markets.
For BDC, AI is both an opportunity and a test of its model. Higher density AI racks and clusters require more sophisticated power and cooling design, tighter operational tolerances, and faster deployment of capacity. Fan’s argument is that a vertically integrated operator with its own IP, R&D, and in-house engineering stands a better chance of meeting those demands reliably, and of capturing more of the value over time.
The company’s patent-backed designs, regional design platform, and insistence on operating its own facilities are not decorative extras. They are the levers BDC hopes will keep hyperscalers anchored to its campuses as AI reshapes how compute is built and consumed.
When asked about how much more does it cost to build DCs for the new AI-driven demand happening, Fan explained that the cost dynamics are shaped by two major factors. “I cannot simplify this answer.”
On one hand, the classic cost curve continues to trend downward thanks to economies of scale and optimizations across the supply chain, benefiting not just his company but the wider industry. On the other hand, adapting data centers to meet the unique requirements of AI workloads definitely introduces new expenses.
For instance, “designing facilities to be more flexible and fungible for AI applications requires investment in advanced cooling technologies – moving away from traditional CPU-based air cooling toward liquid cooling systems that channel coolant directly through chips. This shift is essential for supporting high-performance AI GPU architectures but comes with added costs.”
Pressed further on the scale of these increases, Fan noted that while the exact figures vary depending on application and hyperscale requirements, a reasonable estimate would be a 15% to 20% rise in costs compared to pre-AI era designs. This reflects the balance between ongoing efficiencies in traditional infrastructure and the premium required to build AI-ready data centers.
But whatever the market needs, Fan says, “with speed in our DNA”, BDC, which can build a data centre in under 24 months, is there to deliver and with Malaysia playing a very important role.
(Except for the headline, this story has not been edited by PostX News and is published from a syndicated feed.)
