Every headline about the AI boom seems to feature the same handful of companies: Nvidia, Microsoft, OpenAI, maybe a chip fab in Taiwan. Thailand rarely makes that list. It has no Nvidia, no hyperscale cloud giant, no household-name AI lab. And yet, according to a Thai investment bank, that might not matter as much as it sounds — because the AI boom isn’t just a software story. It’s an infrastructure story, and infrastructure needs to be built somewhere, cooled somewhere, and powered by somebody’s electrical grid.
On September 14, Suphaphong Iamkhongek, an equity research analyst at Kiatnakin Phatra Securities (known locally by its ticker-style abbreviation KKP, one of Thailand’s larger investment banks), laid out a case for why Thailand could be an unlikely winner from the AI data center gold rush — not by building the brains of AI, but by building the plumbing around it. For anyone trying to figure out where the real, less-crowded money is in AI right now — whether you’re a freelancer, an investor, or just someone trying to understand where global supply chains are shifting — this is worth understanding in detail.
The Problem Nobody Wants Living Next Door
Data centers — the warehouses full of servers that actually run AI models — have quietly become one of the most contested pieces of real estate on the planet. Communities from Virginia to Ireland have pushed back against new builds, not because they dislike AI, but because a single large data center can strain a local power grid the way a new small city would.
The scale of the problem is almost hard to process. According to estimates from Bank of America Global Research cited in the KKP briefing, power consumption per server rack — essentially, one column of stacked computing equipment — is projected to rocket from 121 kilowatts in 2024 to 177 kilowatts in 2026, then to 646 kilowatts in 2027, before crossing 1,535 kilowatts from 2028 onward. That’s roughly a twelve-fold increase in power draw per rack in under four years. “What we call massive AI power use today doesn’t even compare to what it will be,” Suphaphong said. “It could be ten times, or a hundred times more.”
More power means more heat, and heat is where old infrastructure starts to fail. Traditional copper wiring, the kind that has carried electricity through buildings for a century, isn’t stable enough or efficient enough at these new power densities — it loses performance as it heats up. That’s pushing the entire data center industry toward two newer technologies: 800VDC power systems (an electrical architecture that moves power at 800 volts direct current rather than the lower-voltage alternating current used in conventional buildings, allowing far more power to move through the same physical space with less energy lost as heat) and optical connectivity, which uses light through fiber-optic cable instead of electrical signals through copper wire to move data and, increasingly, power itself.
This is the pivot point of the whole argument: somebody has to manufacture the guts of this new system — the power converters, the cooling systems, the racks themselves. And that “somebody” doesn’t have to be a country with its own chip designers.
Thailand’s Actual Job in the AI Supply Chain
Suphaphong was blunt about where Thailand stands. “Thailand may not have chips like Nvidia, software like Microsoft or Oracle, or cybersecurity like CrowdStrike,” he said, “but Thailand has all the infrastructure the modern world needs.” KKP breaks that infrastructure opportunity into four buckets:
Server Board Power — the power management and conversion systems that sit physically close to the chip itself, regulating the electricity it receives.
PSU and Power Rack — power supply units and rack-level equipment that distribute and convert electricity for entire server clusters.
Cooling — the systems, from liquid cooling loops to advanced air handling, that keep increasingly hot server racks from overheating.
แนะนำ
Facility Infrastructure — the physical building systems, including high-voltage wiring like 800VDC, that a data center needs to even accept this level of power in the first place.
None of these require Thailand to design a chip. They require precision manufacturing, electrical engineering, and supply chain reliability — areas where Thailand has decades of experience through its existing electronics and auto-parts industries. Bank of America Global Research estimated the global market tied to this kind of infrastructure at roughly $7.9 billion in 2025, rising to $12.3 billion in 2026, and then to $21.8 billion within one to three years. “That means the benefit Thailand could capture will roughly double within the next one to two years,” Suphaphong said.
The Uncomfortable Part: Thailand Is Still at the Bottom of the Value Chain
Here’s where the optimism needs a reality check, and it’s a distinction that matters for anyone thinking about where the real profit sits. In global manufacturing, “downstream” activity means assembly and production — taking components and putting them together. “Upstream” activity means design and high-value engineering — the work that captures the biggest margins. Right now, Thailand sits almost entirely downstream.
Companies like Analog Devices (a US semiconductor maker specializing in chips that manage electrical signals and power), Lumentum (which makes optical and laser components used in data networking), Seagate, and Western Digital (both major hard drive and data storage manufacturers) have all expanded manufacturing operations in Thailand. That’s real investment and real economic activity. But it hasn’t translated into a proportional jobs boom, because modern factories increasingly run on automation, not headcount. As Suphaphong put it, somewhat dryly: it’s “still better than nothing” — foreign direct investment (FDI, meaning money that foreign companies invest directly into local operations, as opposed to buying stocks) still deepens Thailand’s role in the global tech supply chain, even if the job numbers don’t match the headline investment figures.
Can Thailand Actually Move Upstream? Analysts Think So — Eventually
There’s one specific signal Suphaphong pointed to as genuinely encouraging: Analog Devices reportedly has plans to expand its Thai operations into more upstream production processes, rather than staying purely in assembly. If that expands, it would suggest global chipmakers see Thailand as capable of more sophisticated work, not just cheap labor and floor space.
Thailand’s case for moving upstream rests on a few real advantages: an existing electronics supply chain built over decades, reasonably strong power infrastructure, a geographically central location within Southeast Asia, and — perhaps most usefully — the ability to watch and learn from neighboring countries that started this climb earlier. Suphaphong’s own timeline was candid about the pace: “I believe Thailand can move further into upstream industry, but not soon. This is something Thailand is moving toward — maybe in the next five to ten years.” He specifically flagged two sub-sectors where US companies already operating downstream in Thailand could help pull the country upward: AI chip manufacturing, and analog chip production — the specialized chips that regulate voltage and are used everywhere from electric vehicles to the power supplies inside computers and hard drives.
แนะนำ
The Same Race Is Happening All Over Southeast Asia — And Thailand Isn’t Winning It Alone
Zoom out from Thailand and the same story is playing out, at different speeds, across the entire region. Vietnam has drawn more than $7 billion in AI data center investment, including a $1 billion joint venture between Create Capital Vietnam and Haimaker.ai to build a nationwide AI-focused data center network — part of a wave of state-backed digital infrastructure spending and new AI-specific regulation that took effect in March 2026. Deloitte has projected Vietnam’s AI market could reach $65 billion by 2035, with $25 billion of that coming from AI data centre infrastructure alone.
Indonesia is playing a different game — betting on the sheer size of its domestic digital economy rather than export manufacturing. Microsoft alone committed $1.7 billion to cloud and AI infrastructure there in 2024, and local operator DCI Indonesia has been scaling from 83 megawatts of capacity in 2023 toward a potential 1,000 megawatts. Regional cloud demand has been compounding fast enough that Indonesia’s cloud market has grown at a 48% annual rate over five years, with roughly 90% of companies now moving workloads to the cloud.
And Thailand itself isn’t standing still on the data center side either, even setting aside the supply chain argument: Bangkok’s active data center building capacity hit 859 megawatts in early 2026, a 148% jump from the previous period, with a total development pipeline of about 2.08 gigawatts. Industry analysts expect the broader Southeast Asian data center market — spanning Indonesia, Malaysia, Singapore, Thailand, Vietnam, and the Philippines — to keep expanding at roughly 20% a year through 2028, with total regional data center investment potentially reaching $30 billion by 2030.
The pattern across all of these countries is nearly identical to Thailand’s: none of them are designing the chips. All of them are racing to build the power, cooling, land, and physical infrastructure the chips need to run. It’s a regional supply-chain land grab happening underneath the AI story most people are paying attention to.
What This Actually Means If You’re Trying to Make Money Around AI
This is where the story stops being abstract macroeconomics and starts being relevant to individual decision-making — for investors, for freelancers, and for anyone building a career or business around the AI wave rather than around AI models themselves.
For investors, the takeaway isn’t “buy an AI stock.” It’s that the picks-and-shovels layer of the AI boom — power management components, cooling systems, industrial electrical equipment, and the Thai-listed suppliers and contract manufacturers feeding into that global $12–22 billion infrastructure market — is a less crowded, less hyped place to look than chasing chip-design companies whose valuations have already priced in years of future growth. It’s the same logic that made semiconductor equipment suppliers, rather than semiconductor companies themselves, one of the quieter winning trades of earlier tech cycles.
For freelancers, consultants, and small business owners — particularly those based in or serving Southeast Asia — the opportunity is more direct. Every foreign company setting up manufacturing or data center operations in Thailand, Vietnam, or Indonesia needs local support that has nothing to do with AI expertise: technical translation and documentation, compliance and permitting consulting, logistics coordination, facilities management, recruiting for specialized electrical and cooling engineering roles, and content or market research explaining these fast-moving local regulatory environments to head offices abroad (which, incidentally, is exactly the kind of explainer content this article itself represents). None of that requires you to understand how a large language model works. It requires understanding how the physical world that supports AI actually gets built.
For anyone thinking about skills rather than income streams, the durable ones aren’t necessarily “prompt engineering” — they’re the unglamorous technical trades this whole boom depends on: industrial electrical work, HVAC and liquid cooling systems, power electronics, and project management for large industrial builds. Demand for those skills is set to keep climbing in this region for years, arguably with less competition than the flooded market for generic “AI consultant” positioning.
The Bottom Line
Thailand’s AI opportunity isn’t the one most people assume when they hear “AI boom” — there’s no local ChatGPT competitor, no chip design house waiting to be discovered. What’s actually happening is quieter and, in some ways, more durable: a multi-year buildout of the unglamorous physical infrastructure — power, cooling, wiring, assembly — that every AI data center on earth needs regardless of which company’s model is running inside it. That buildout is happening in Thailand, and in parallel across Vietnam and Indonesia, whether or not the country ever designs its own chip. The window analysts are describing is measured in years, not weeks — this is closer to positioning for a structural shift than chasing a headline. Whether you’re evaluating a stock, a business idea, or your own next skill to learn, the practical move is the same: stop looking only at who’s building the AI, and start paying attention to who’s building the room it lives in.
Key Takeaways
- Thailand isn’t designing AI chips, but it’s positioned to profit from the physical infrastructure AI data centers need: power systems, cooling, and facility buildout.
- Data center power demand per server rack could rise roughly twelvefold between 2024 and 2028, forcing a shift to new technologies like 800VDC power and optical connectivity.
- The global market for this AI-adjacent infrastructure was estimated at $7.9 billion in 2025 and could reach $21.8 billion within one to three years.
- Thailand currently sits downstream (assembly and manufacturing) in the supply chain rather than upstream (design and high-margin engineering), though analysts see a 5-10 year path to moving up.
- Vietnam and Indonesia are running the same playbook at different speeds, making Southeast Asia’s data center buildout a regional trend, not just a Thai one.
Frequently Asked Questions
Q: Can you really make money from Thailand’s AI boom without working in AI directly?
A: Yes — the opportunity described here is in the physical supply chain around AI (power, cooling, manufacturing), not in building AI models or software, so it’s accessible to investors, engineers, and service providers with no AI background.
Q: What is KKP and why does its analysis matter?
A: KKP, short for Kiatnakin Phatra Securities, is one of Thailand’s established investment banks, and its equity research team regularly publishes analysis used by both institutional and retail investors to gauge Thai market opportunities.
Q: Why are data centers becoming so power-hungry?
A: AI workloads require increasingly dense clusters of high-performance chips, and Bank of America Global Research estimates power use per server rack could rise roughly twelvefold between 2024 and 2028 as AI computing scales up.
Q: What is 800VDC and why does it matter for AI infrastructure?
A: It’s an 800-volt direct current power system that can move far more electricity through the same wiring with less energy lost as heat, which traditional copper-based systems struggle to handle at AI-scale power loads.
Q: Does Thailand manufacture AI chips like Nvidia does?
A: No — Thailand has no major chip designer of its own, but it hosts manufacturing and assembly operations for companies like Analog Devices, Lumentum, Seagate, and Western Digital that supply components used in AI and data infrastructure.
Q: What’s the difference between “upstream” and “downstream” in this context?
A: Downstream refers to assembly and manufacturing work, which is lower-margin and where Thailand currently sits; upstream refers to design and high-value engineering, which captures more profit and is where analysts hope Thailand can eventually move.
Q: Is Thailand’s AI infrastructure boom creating a lot of new jobs?
A: Not as many as the investment figures might suggest, since modern automated factories require less manual labor per dollar invested than older manufacturing models.
Q: How does Thailand compare to Vietnam and Indonesia in this AI infrastructure race?
A: All three are competing for the same kind of data center and manufacturing investment, with Vietnam drawing over $7 billion in AI data center investment and Indonesia leaning on strong domestic cloud demand, while Thailand leans on its existing electronics manufacturing base.
Q: What are practical ways an individual or freelancer could benefit from this trend?
A: Opportunities include technical translation and compliance consulting for foreign companies entering the region, project support roles in electrical and cooling engineering, and content or research work explaining local regulatory and market conditions to overseas businesses.
Q: Is investing in Thai infrastructure-supply companies a way to get AI exposure?
A: It’s one way analysts suggest getting exposure to AI-driven demand without buying into already-expensive AI software or chip-design companies, though as with any investment, it carries market and company-specific risk and isn’t guaranteed.
Q: How soon could Thailand actually move into higher-value (“upstream”) AI manufacturing?
A: The KKP analyst quoted in this piece estimated a timeline of roughly five to ten years, calling it a direction Thailand is heading in rather than something imminent.
Q: Where is most of this AI infrastructure investment in Thailand actually going right now?
A: Primarily into downstream manufacturing and assembly by foreign companies such as Analog Devices, Lumentum, Seagate, and Western Digital, rather than into chip design or high-value R&D work.
