Armenia’s Missing Middle: Why Science Investment Isn’t Enough

Listen to the AI generated audio article. 

Armenia is investing in the visible parts of a modern research and innovation system, including laboratories, computing capacity, grants and new institutions. But the less visible infrastructure is harder to build. Researchers need supervisors and career paths. Laboratories need engineers and technicians. Companies need the capacity to use research. Public institutions need to become sophisticated customers. And someone has to connect all of these pieces.

That missing middle may be the biggest test facing Armenia’s new science and technology ambitions.

Nearly three years ago, in EVN Report, I asked whether Armenia was building a system in which research and development (R&D) could contribute meaningfully to innovation. In October 2023, the answer was sobering: Armenia had gifted scientists, established institutes, and a dynamic technology sector, but neither a coherent, adequately financed R&D ecosystem nor a framework to make those assets reinforce one another.

Much has changed. Salaries at public research institutions and competitive grant funding have risen, and new laboratories and computing capacity have come online. On January 29, 2026, the government adopted the Science Sector Development Strategic Program for 2026–2030. On May 29, the Ministry of High-Tech Industry published a draft High-Technology Sector Development Strategy for 2026–2033. Together, they could shape Armenia’s scientific and technological landscape for decades.

The question has therefore changed. Armenia can now articulate much of the future it envisions. The test is whether the state can integrate the emerging building blocks into a system that is capable of developing, adapting, and diffusing knowledge; creating economic and public value; and giving talented people a reason to build their lives in Armenia. It is also a test of staying power: whether the state will honor its promises beyond a single budget or political cycle through sustained financing, a consistent strategic vision and patient investment in human capital.

The System and Its Three Clocks

The science strategy begins where Armenian science has been starved of for years: salaries, grants, careers and public laboratories. The draft high-tech strategy focuses on firms, applied R&D, investment, standards and commercialization. These priorities are interconnected but the strategies do not yet identify a clearly empowered mechanism for coordinating the dependencies among research institutions, universities, firms, regulators, investors and public users.

A national innovation system is not a catalog of organizations with science, technology or innovation in their names. It is the set of institutions, incentives and relationships through which knowledge is created or acquired, tested, adopted and diffused. A company’s problem can become a research question; a laboratory result can be engineered and validated; an instrument developed in one institution can become useful elsewhere. Each stage should feed into the next.

This is where Armenia’s missing middle becomes important. The country does not lack ambition, assets or diagnosis. What it lacks are the functions, rules and relationships that connect high-level strategy and individual projects to sustained national capability. That middle includes career pathways, shared-facility governance, research engineering, technology transfer, regulation, procurement and evaluation.

Building it is difficult because Armenia’s ambitions operate on three clocks that are not synchronized. The infrastructure clock measures the time required to build laboratories and computing systems. The human-capital clock measures the longer process of developing researchers, engineers, technicians and scientific leaders. The institutional-learning clock measures the time organizations need to develop trust, shared standards, collaboration, and the ability to learn. Infrastructure is the most visible and politically legible of the three, but the other two determine whether it becomes productive.

This is why a national capability strategy cannot be reduced to infrastructure targets or funding levels. It also requires a clear picture of the people and institutional capacity needed to make those investments work; how many research leaders, postdoctoral researchers, supervisors, engineers, specialists and program managers are needed, how many exist and how many can realistically be developed, attracted and retained.

Without that account, institutional expansion can be mistaken for capability growth. A new laboratory may compete with an established one for the same technical specialist. A new grant program may increase demand for supervisors without enlarging the pool of qualified mentors. Strategic concentration can create critical mass. But competition for the same scarce resources enlarges the institutional footprint without expanding the country’s actual capacity.

AI offers the clearest illustration. Armenia has assembled or announced substantial computing capacity through Yerevan State University, Eleveight AI and Firebird. Yet infrastructure is expanding faster than the research teams, doctoral supervisors, software engineers, infrastructure specialists and sophisticated users needed to exploit it. Compute can be acquired much faster than the people and institutions needed to use it well can be developed. In AI, that gap is especially costly because hardware, models and technical standards advance rapidly. Capability that develops too slowly may arrive after the strategic value of the original infrastructure has already diminished.

The problem extends beyond research. In Armstat’s 2023 ICT survey of 1,974 firms, more than 60% reported adopting none of the modern technologies examined. Among manufacturing establishments with at least 10 employees, reported AI use was below 1%. The challenge, then, is not simply giving Armenia access to new technology, but building the capacity of researchers, firms and public institutions to use it.

Thus, Armenia needs a shared operating logic across government, research institutions and business: which capabilities should be concentrated and which distributed; how new institutions and firms will expand rather than simply compete for scarce talent; how expensive infrastructure will be accessed, operated and maintained; how companies and public bodies will articulate demand, test and adopt solutions; and who is responsible when coordination breaks down between research, engineering, regulation, finance, procurement and use.

Oases in a Dispersed System

The science strategy identifies 54 organizations and institutions receiving public base science financing in 2025, which differ sharply in affiliation, size and scientific strength. The strategy further reports 13,743 Web of Science-indexed publications affiliated with Armenian research organizations and universities during 2015–2024. Excluding publications with more than 50 coauthors, the figure falls to 11,066. Within this subset, Armenian researchers collaborated with scientists or research centers in 167 countries.

Armenia’s participation in European research programs has also grown. Armenian organizations received over €4 million through 35 funded Horizon 2020 proposals. At the Horizon Dashboard cutoff of October 17, 2025, Armenia had 26 signed Horizon Europe grants involving 28 Armenian organizations and €8.1 million in EU contributions.The question is whether these projects leave behind stronger research teams and management capacity, and whether they broaden Armenia’s participation in frontier and industry-oriented research.

These figures reject the caricature of Armenian science as uniformly weak or isolated. They demonstrate real capability, often achieved under unfavorable conditions. Yet exceptional laboratories can give a misleading impression. The success of particular groups does not mean that the wider system functions effectively.

These are Armenia’s scientific oases: proof that excellence is possible, but not yet reliably reproduced or connected. International networks preserve standards and keep fields alive. Yet a scientist may be embedded in a European network while remaining distant from an Armenian university, hospital or company. Strong international ties can coexist with weak domestic circulation of people, knowledge and infrastructure.

This asymmetry has historical roots. The Soviet system linked teaching, research, engineering and production through a unified command economy. Armenia inherited parts of the scientific apparatus but not the wider system of commissioning, production and procurement that supported it. Teaching remained separated from research; public research, from firms and users. A 2020 European Commission review documented a persistent divide between higher education and research, asking whether they were “two worlds or one?” It found limited cooperation between the research system and the broader economy.

From Knowledge to Use

Modern innovation doesn’t follow a one-way pipeline. Research and innovation serve different purposes and not all research should target commercialization. Investigator-led science can produce methods, skills, and discoveries whose value emerges years later. But when research is intended for practical use, moving from knowledge to application requires more than funding. Firms and public institutions must be able to identify problems, assess solutions, test them and adopt what works.

Innovation therefore depends on feedback between researchers, engineers, regulators and users. A company or public agency may define a problem that becomes a research question; a discovery may create an opportunity no market had anticipated. The Oslo Manual defines innovation as a new or improved product or process introduced into use. R&D spending alone cannot produce that outcome. As the World Bank’s “innovation paradox” suggests, firms also need complementary capabilities, while governments need the capacity to design and implement effective innovation policy.

This is not an argument for postponing funding. Predictable finance and absorptive capability are complements: prolonged underfunding destroys the teams and routines that make additional investment productive. Armenia must therefore increase financing while building capabilities that persist.

Armenia’s technology-sector data show why the distinction matters. According to the draft high-tech strategy, the tax-record cohort of entities operating for at least one day during the reporting year rose from 4,640 in 2021 to 13,201 in 2024; 9,482 were active at the end of 2024. The population was predominantly small by employment and young by operating history.

Its composition changed sharply. Among active entities, the share whose principal activity was IT services rose from 63% to 86%, while high-technology manufacturing’s share fell from 24 to 10%. Scientific research and experimental development represented about 2%.

This is not a census of science-based companies. A diagnostics or advanced-manufacturing firm may conduct R&D while classified by what it sells; or it may collaborate with an R&D unit at a research institute or university rather than maintaining an in-house R&D team.The evidence therefore clearly shows rapid, service-heavy business formation, but it does not establish whether research-intensive business activity expanded at a comparable pace. Armenia still lacks comprehensive data on private R&D investment and personnel.

The high-tech strategy draft illustrates the same point from another angle. It examines 14 fields and selects six priorities: artificial intelligence; data science and machine learning; software development; bioengineering and medical technologies; advanced engineering solutions; and sustainable technologies.

If Armenia wants a research-intensive high-tech sector in these fields, not merely more IT services, it must address the capability failures behind the innovation paradox while increasing R&D. More spending will not reliably produce innovation if firms and public institutions cannot absorb research, engineers cannot turn it into reliable systems, regulators cannot validate it and no capable customer will adopt it.

The draft’s talent assessment shows why one policy package cannot serve all six. Data science, AI and software rank the highest in terms of talent supply; bioengineering and medical technologies rank relatively lower; sustainable technologies and advanced engineering much lower. Their needs also differ. AI and software need product firms, demanding customers and routes to scale; bioengineering and medical technologies need clinical and regulatory pathways; advanced engineering and sustainable technologies need teams, test facilities, industrial partners and patient finance. Agile policy must identify each field’s principal constraint, fund the missing capability alongside research, and revise instruments as evidence changes. Otherwise, generic grants and tax incentives will reinforce the fields best able to absorb them.

The scale and design of R&D investment are inseparable. The science strategy puts state funding for scientific and science-and-technology activities at AMD 37.8 billion in 2026, or 0.31% of GDP, down from AMD 40.2 billion, or 0.36% of GDP for 2025. A separate AMD 7.4 billion envelope for priority measures is conditional on additional budget resources or a legally authorized increase in the deficit ceiling. The action plan targets total R&D expenditure of 1% of GDP by 2030—0.75% public and 0.25% private.

So far, the money has not kept pace with the promises. Reaching 0.75% of GDP by 2030 would require public funding for science to rise by roughly a quarter every year relative to GDP. Instead, the conditional AMD 7.4 billion has not been allocated, and the draft 2027 budget, approved by the government on October 1, raises overall funding for science and technology activities by only 5%, not enough to reach the target of 2030. Budgets for competitive research grants and targeted research for state needs remain unchanged, leaving little room for new competitions or new research teams, even as the allocation for building the Academic City rises by more than 36%. The infrastructure clock is once again running ahead of the human-capital clock.

People Set the Speed

Armenia has made a welcome, though insufficient, correction after years in which scientific careers had become economically precarious. For attested scientific personnel working full-time in state-budget base-financing programs, the strategy reports a 2021 average gross monthly salary of AMD 85,800 for a junior researcher and AMD 138,100 for a principal researcher. The corresponding 2025 minimum monthly rates were AMD 210,000 and AMD 552,900. The measures are not identical—averages are being compared with minimum rates—but the correction is substantial.

The reported headcount of scientific staff participating in base-funded programs rose from 3,139 in 2021 to 3,761 in 2024. The number of scientists under 40 and under 30 also increased. The timing is consistent with salary and grant reforms improving recruitment or retention, but it does not establish causation or reveal how much of the increase reflects classification changes. The 2030 targets are 6,000 publicly financed FTE researchers and 8,000 researchers in total when the private sector is included.

The objective is ambitious, but so is the succession problem. In 2024, 29.5% of scientific staff were older than 65, while 214 of 442 scientific leaders, 48.4%, were above pension age. Senior scientists are not the problem. The risk is the absence of strong overlapping generations capable of transferring tacit knowledge and eventually assuming leadership.

The doctoral pipeline cannot yet provide succession at the required scale. The strategy juxtaposes 897 state-funded doctoral places allocated during 2017–2021 with 441 Candidate of Sciences dissertations defended by doctoral students during 2020–2024. Moreover, 350 people earned Candidate of Sciences certificates in 2023–2024, yet only 107 participated in state-financed research programs in 2025. This does not prove that the others left science, but it signals limited visible absorption into publicly financed research.

The action plan seeks doctoral enrollment of 800, up from 553; 400 current beneficiaries of doctoral research grants, up from 110; 1,500 recipients of scientific degrees, 1,000 participants in overseas training, and 150 Global Talent beneficiaries (up from 25) during 2026–2030. These are substantial commitments, but most measure participation and throughput rather than durable careers. How many complete on time, obtain postdoctoral positions, form independent groups and remain in Armenia after three or five years?

The transition after the doctorate remains weakly institutionalized. The strategy recognizes the need for postdoctoral opportunities, but its action plan contains no separately identified postdoctoral measure, annual target or budget line. International recruitment faces the same test. Visits are valuable, but a field is built when researchers receive multi-year support, form teams, supervise students and remain long enough for methods and leadership to take root locally.

People are the slowest component of any infrastructure promise, and the source of its value. A supercomputer can be bought within a budget cycle, but an independent scientist takes shape over years of apprenticeship, failure, and judgment. Leadership, staffing, and career pipelines must therefore be planned and funded before or alongside the infrastructure, not treated as questions to answer after a facility opens.

The Hard Part Begins After the Strategies

Armenia’s strategies are notably candid about the country’s structural problems. That candor has strategic value: the principal weaknesses of the national innovation system are no longer hidden. But diagnosis is only a starting point; it is not evidence that implementation is happening.

The head of the state body that coordinates higher education and science made the same point in resigning. Sargis Hayotsyan, who has chaired the Higher Education and Science Committee since 2020, stepped down at his own request on September 29. Two days later, he wrote that he still believed in the government’s reforms, but that reforms of this scale cannot succeed through the right policies and concepts alone. They require “consistency at all levels, a clear distribution of institutional responsibility, effective, fast and flexible coordination, concentration of thought and effort, and long-term continuity.” In his professional assessment, he could no longer ensure the pace, consistency, and accountability that proper implementation required, given the current implementation environment and the tools available to him.

The weaknesses form a causal hierarchy. At the foundation is a shortage of complete research and engineering teams, credible career paths, and experienced organizational leadership. Above that are weak interfaces among research, engineering, regulation, procurement, production, and use. At the system level, fragmented mandates, incomplete data, and limited implementation capacity make the underlying problems harder to coordinate and correct.

The answer is not to place science, higher education, industry, procurement, health and regulation in one bureaucracy. Their mandates are different for good reason. What Armenia lacks is an operating layer that allows these institutions to work together: a shared capability map and implementation timetable; named responsibility for critical interfaces; recurring finance and access rules for facilities; mechanisms connecting research with validation and first-customer purchase; and outcome measures showing what remains after public support ends.

This also requires professional capacity within the state: enough qualified staff, analytical resources, and authority to track dependencies, compare instruments, identify failed assumptions, and revise programs while implementation is under way. Science and technology will not wait for Armenia’s strategy cycles. A government that diagnoses today’s problems but cannot revise its instruments until 2030 or 2033 will still be governing from the past.

A strategy is not a prophecy to defend. It is a disciplined hypothesis about how institutions can produce change. Armenia’s national innovation system will not be built by the ambition of its documents, the visibility of its infrastructure, or the number of institutions it creates. It will be built through the continuity and professionalism required to fill Armenia’s missing middle with durable, systematic, and institutional connections.

That is where a doctorate becomes a career; senior scientists prepare successors; a new institution creates rather than redistributes talent; a laboratory becomes part of a shared network; research encounters engineering and validation; a public institution becomes a demanding customer; expenditure becomes durable capability; and evidence changes the next program.

A functioning national innovation system cannot by itself guarantee transformative innovation or long-term growth. But without these connections, Armenia’s rising investment in science and technology is unlikely to compound into innovation, productivity, and public value. The country may acquire stronger scientific oases, impressive infrastructure, and ambitious institutions while the system connecting them remains unfinished. Under those conditions, transformative growth is difficult to imagine.

Creative Tech