The Department of Science and Technology formally launched the National Artificial Intelligence Center for Research and Innovation, NAICRI, on February 26 at the Manila Hotel, consolidating what had been years of fragmented AI research funding, computing infrastructure, and capacity-building programs into a single, permanent national institution. The launch drew senior government officials, members of Congress, academe, industry representatives, and international development partners, and DepEd publicly welcomed the launch the same week, specifically citing responsible AI integration into education as one reason the department was watching closely.
NAICRI’s stated mandate is to serve as the Philippines’ institutional anchor for AI research priorities, partnerships, and standards, coordinating what had previously been separate, project-by-project initiatives across government, academic, and industry lines. It’s built around four strategic pillars: national AI computing infrastructure, AI talent and research capability development, governance and strategic coordination, and what DOST is calling regional innovation and inclusive growth, explicit language aimed at the same Metro Manila-concentration problem that shows up in nearly every other assessment of the Philippine tech ecosystem, funding, talent, and now, potentially, compute access.
The infrastructure NAICRI is built on top of already exists and has a real operating history. COARE, the national supercomputing facility now folded under NAICRI’s umbrella, has powered projects including FASSSTER, the country’s pandemic surveillance and forecasting system, rice genome sequencing work done jointly with the International Rice Research Institute, and climate modeling and hazard mapping used for disaster preparedness. DOST reports COARE’s GPU utilization has risen 54 percent since 2020, a concrete sign of rising demand for the compute capacity that already existed before NAICRI gave it a formal institutional home. PREGINET, the country’s dedicated research and education network, connects Philippine academic, government, and research institutions to global research networks, the kind of connectivity infrastructure that matters enormously for AI research but rarely gets any public attention at all.
What NAICRI adds isn’t new hardware, at least not yet, it’s institutional permanence and a coordination mandate that COARE and PREGINET never had on their own. Under its initial roadmap, DOST plans to run national surveys within 30 days of launch to build what it’s calling an AI compute country profile, essentially an inventory of who in the country, which universities, which government agencies, which private companies, actually has access to meaningful AI compute today and where the gaps are, followed by a training catalogue meant to help researchers and institutions in the regions actually make use of shared computing facilities rather than needing to independently discover or negotiate access to them.
That regional-access framing is the detail worth reading against the much larger, and much better-funded, AI infrastructure story already unfolding a short distance away. A separate, US-backed plan to build a roughly $10 billion AI supply-chain hub near Clark Freeport Zone, data centers, semiconductor facilities, green-energy and logistics infrastructure, has drawn far more international coverage this year, and for good reason: it’s an order of magnitude larger in dollar terms than anything DOST is funding directly through NAICRI. But that Clark hub is fundamentally a private, foreign-investment-driven infrastructure play, aimed at positioning the Philippines as a trusted node in global AI supply chains for outside capital and outside customers. Whether any of that compute capacity ends up accessible to Filipino researchers, startups, or SMEs at competitive rates, rather than serving hyperscaler and enterprise clients exclusively, was always the open question hanging over that project.
NAICRI is, in effect, the domestic-access counterpart to that question, DOST building the coordination and allocation layer that would let existing Philippine compute infrastructure, and potentially future infrastructure like Clark’s, actually reach researchers and companies outside the handful of institutions that have historically had the connections to access it. A national AI compute country profile is a genuinely useful diagnostic tool precisely because nobody currently has a clear public picture of where the country’s actual compute bottlenecks sit, which university labs are starved for GPU time, which regional SMEs have AI ambitions but no path to the infrastructure needed to pursue them.
The gap between NAICRI’s ambition and its resourcing is real, though, and worth naming plainly rather than glossing over. A 30-day national survey and a training catalogue are planning and access-coordination tools, not new compute capacity. NAICRI’s launch materials don’t include a large new hardware procurement budget attached to the announcement, and COARE’s rising GPU utilization since 2020 is itself evidence that existing capacity is already under real demand pressure before any new access-broadening initiative adds more users into the queue. Consolidating fragmented programs under one institutional roof is a genuine and useful structural fix, years of siloed, project-by-project AI funding is a real problem NAICRI’s four-pillar design directly addresses, but it doesn’t by itself create more GPU hours in the country.
The honest read on NAICRI, six months in, is that it’s necessary institutional plumbing rather than a transformative infrastructure investment on its own terms. Whether it succeeds will depend less on its launch event at the Manila Hotel than on whether the AI compute country profile DOST promised within 30 days actually gets published, whether the training catalogue reaches researchers outside Metro Manila who’ve never had a clear path to shared computing before, and whether NAICRI’s regional-inclusion pillar ends up meaning anything more concrete than the phrase itself.
Share this article