Nguyen Manh Thang
Chief Executive Officer, AgileTech Vietnam
- Published
- Read time
- 10 min
In short
Vietnam's software industry divides into four segments a buyer should understand before shortlisting: global technology companies running large engineering sites in the country, led by Samsung's R&D operation in Hanoi; large domestic firms with tens of thousands of engineers, led by FPT, which serve enterprise clients worldwide; product companies and startups like VNG, MoMo and Sky Mavis that prove the market's product engineering depth; and hundreds of mid-size specialist firms of 50 to 1,000 engineers where most foreign buyers actually find their fit. Published top-ten lists are largely pay-to-play directories and should be treated as advertising; a real shortlist starts from your work's shape, then filters by segment, domain evidence and the diligence questions that separate partners.
Search for the top software companies in Vietnam and you will find a dozen near-identical listicles, each ranking firms that, on inspection, sponsor the directory doing the ranking. This page is a different tool: a map of the industry's actual structure, the four segments that matter, the names that anchor each, and the honest logic for matching a buyer to a segment, written by a Hanoi firm that competes in one of them and says so.
The map matters because the segments serve different buyers badly or well. A startup that engages a 30,000-person enterprise firm gets a junior team and a distant account manager; an enterprise that hires a 40-person boutique for a five-year core banking program takes concentration risk it should not. Most selection mistakes in this market are segment mistakes before they are vendor mistakes.
This page pairs with three neighbors: the offshore Vietnam guide covers why teams locate here at all, the rate benchmark prices the geographies, and the partner evaluation guide covers the vendor-level diligence that follows once the segment is right.
Key takeaways
- The industry is four segments, not a ranked list: global giants' engineering sites, large domestic firms, product companies, and mid-size specialists, and each fits a different kind of buyer.
- The scale anchors are real: Samsung runs one of its largest R&D sites in Hanoi, FPT passed a billion dollars of annual global IT services revenue, and the market employs over half a million engineers.
- Product companies are the depth signal: VNG, MoMo, Sky Mavis and the games industry demonstrate that the market builds and operates its own products at scale, not just other people's tickets.
- Most foreign buyers fit the mid-size specialist segment: 50 to 1,000 engineers, senior attention on your account, domain depth in specific verticals, and rates below the large firms' enterprise pricing.
- Published top-ten lists are advertising: directory rankings are substantially pay-to-play, so treat them as a phone book, not an evaluation, and do your own diligence.
- Shortlist from the work backward: the segment that fits your engagement size, then domain evidence, then the partner-level diligence questions, attrition, tenure, ownership culture, that actually predict outcomes.
The landscape in one view
Vietnam's software industry employs over half a million engineers across a structure that formed in layers. The oldest layer is enterprise services: firms founded in the 1990s and 2000s to serve Japanese, Korean and later Western enterprises, which grew into today's large domestic companies. The second layer is the global capacity build-out: multinational technology companies opening their own engineering sites to hire the same talent directly. The third is the product layer: domestic startups and product companies, funded by the region's venture capital, building for Vietnamese and Southeast Asian consumers. The fourth, and the largest by firm count, is the mid-size specialist tier serving foreign clients directly.
The segments compete for the same engineers, which is exactly why the map matters to buyers. Salary and career expectations are set market-wide: a strong engineer in Ho Chi Minh City weighs a global firm's brand, a product company's equity, and a services firm's client variety, and the firms that win talent are the ones whose economics let them pay competitively. When a vendor's rates undercut the market dramatically, the discount is coming from somewhere, usually seniority mix or turnover, and understanding the segment economics is how buyers read the discount.
Geography concentrates the market in three cities. Hanoi holds the government-adjacent enterprises, Samsung's R&D center, and a deep pool shaped by the capital's technical universities. Ho Chi Minh City is the commercial center, the startup ecosystem's home, and the largest talent pool overall. Da Nang, smaller and newer, grew as the deliberate second-site choice for firms hedging big-city salary inflation and attrition. Most sizable vendors now operate in at least two of the three, which matters practically: a partner with multi-city delivery can hire around any one city's market heat.
One structural fact frames every conversation with every vendor: this is a young market in age and a maturing one in capability. The median engineer is under thirty; the senior architect layer is thinner than in India or Poland, as the offshore guide covers honestly; and the capability curve has been climbing visibly for a decade, from maintenance tickets to owned products. Buyers who calibrate to where the market actually is, rather than to either the marketing or the stereotype, make better segment choices.
The four segments, defined
- Global engineering sites
- Multinationals' own operations in Vietnam: Samsung R&D in Hanoi, plus sites run by global technology and consulting firms. Employers, not vendors.
- Large domestic firms
- Vietnamese enterprises with thousands to tens of thousands of engineers, led by FPT. Enterprise clients, global delivery, full-service portfolios.
- Product companies
- Firms whose product is their own: VNG, MoMo, Sky Mavis, the games studios. Proof of the market's product engineering depth, and competitors for its talent.
- Mid-size specialists
- The hundreds of firms from 50 to 1,000 engineers serving foreign clients directly, usually with vertical or stack specialization. Where most foreign buyers fit.
The global giants: what their presence tells a buyer
Samsung anchors the segment and the signal. Its Hanoi R&D center, opened in its current campus form in 2022, employs thousands of engineers and stands among the company's largest research sites outside Korea, working on mobile software, networks and devices. Around it sit LG's vehicle software operation, Panasonic and Bosch engineering sites, and the delivery centers that global consultancies and technology firms run to serve their own clients from Vietnamese capacity. None of these are vendors a typical buyer can hire; all of them are evidence about the market.
The evidence reads three ways. First, talent validation: multinationals audit markets hard before committing R&D sites, and their sustained expansion is a stronger signal about engineering quality than any directory ranking. Second, training effect: these sites run global engineering practices, code review culture, security process, large-system discipline, and their alumni carry those practices into the domestic industry, seeding the specialist tier with people trained to multinational standards. Third, wage pressure: the giants pay top of market, which pulls the whole salary curve upward, one reason rates rise 10 to 15 percent annually in hot segments.
For buyers, the practical use of this segment is calibration and, occasionally, competition. Calibration: when a vendor claims its engineers meet global standards, alumni of the global sites on the team are checkable evidence. Competition: a buyer opening its own entity, the path covered in the offshore team setup guide, is entering the same talent market these employers define, and should price its offer against them rather than against services-firm salaries.
The segment also explains a pattern buyers notice in interviews: Vietnamese engineers who have never worked for a services firm at all, moving between product companies and multinational sites. That career path barely existed fifteen years ago. Its normality now is the clearest single indicator of how far the market has moved from its subcontracting origins, and why treating Vietnam as a ticket-execution destination undervalues what is available here.
The large domestic firms: FPT and the enterprise tier
FPT is the segment's defining company and one of the industry's founding stories. Started in 1988, its technology arm passed one billion dollars in annual revenue from global IT services in 2023, employs tens of thousands of engineers, and operates delivery centers across Vietnam plus offices in dozens of countries. Its client list is the global enterprise tier, automotive, banking, healthcare, aviation, and its portfolio spans the full services range from legacy modernization to AI programs, increasingly through acquisitions in the US, Europe and Japan.
Around and behind FPT sit the other large domestic players: CMC, the second-largest domestic technology group; Viettel's software arms, backed by the military-run telecom giant; VNPT's technology units; and a band of firms in the one-to-five-thousand-engineer range, names like TMA Solutions, KMS Technology, NashTech and Rikkeisoft, that grew on Japanese and Western services demand. The tier's shared profile: process maturity certifications, multi-hundred-person delivery capability, established legal and security infrastructure, and account structures built for enterprise procurement.
The tier's fit logic is straightforward on both sides. It fits enterprise buyers running large, long programs: a hundred-person modernization, a multi-year managed service, work that needs certified process, audit trails, business continuity plans and the balance sheet to guarantee them. It fits badly the buyer this blog most often serves: the product company that needs a senior eight-person team with direct communication, for whom the enterprise tier's account structure adds relay hops and its best engineers are usually deployed on accounts a hundred times larger.
The honest trade inside the tier is scale against attention. The large firm can staff fifty engineers in a quarter, absorb attrition invisibly, and survive any single client's departure, real advantages at enterprise scale. The cost is that a small engagement is a small account: junior-heavy staffing, manager-mediated communication, and process designed for programs rather than products. Neither side of the trade is a defect; matching your engagement size to the tier's economics is the whole game.
The enterprise tier's anchors
The product companies: the market's depth proof
The product segment is the part of the landscape most buyers never engage commercially and should study anyway, because it answers the capability question directly. VNG, the games and messaging company behind Zalo, Vietnam's dominant chat platform, operates consumer infrastructure at a hundred-million-user scale. MoMo runs one of Southeast Asia's largest payment and super-app platforms. Sky Mavis built Axie Infinity, briefly the world's most valuable blockchain game, from Ho Chi Minh City. The games industry ships global hits from studios across both major cities, a lineage running back to Flappy Bird.
What the segment proves is precisely what ticket-outsourcing stereotypes deny: that the market designs, builds and operates its own products, with the product management, growth engineering, reliability practice and iteration speed that implies. The super app case studies elsewhere on this blog, MoMo's platform anatomy among them, exist because these companies' engineering is worth studying on its own terms, not as a curiosity.
For buyers, the segment matters in two practical ways. It is the strongest talent competitor: product companies offer equity, ownership and consumer-scale problems, and they set the bar a services vendor's retention economics must clear, which is worth remembering when a vendor's proposed team looks suspiciously cheap. And it is a hiring pool signal: engineers with product-company backgrounds on a vendor's team, like global-site alumni, are checkable evidence of practices beyond ticket execution.
The segment's existence also carries a strategic implication for the market's future that buyers planning five-year commitments should weigh. Markets that develop product layers, as India's did, keep climbing the value curve: rates rise, but so does the ceiling of what can be bought. A buyer building a long-run team in Vietnam is buying into that trajectory, senior product engineers will be more available in five years, not less, which is the opposite of betting on a static labor-cost arbitrage.
The mid-size specialists: where most buyers actually fit
The segment with the most firms and the least directory visibility is the one most foreign buyers belong in: specialists from roughly 50 to 1,000 engineers, serving foreign clients directly, usually with a recognizable center of gravity, a vertical like fintech, healthcare or logistics, a stack like mobile or data engineering, or a client geography like Japan, Australia or the US. AgileTech, the company publishing this guide, is one of them, and the segment description that follows is also a self-description, offered with that disclosure.
The segment's economics explain its fit. A firm of a few hundred engineers lives or dies on a portfolio of ten to forty client relationships, which changes the attention arithmetic completely: your eight-person team is a meaningful account, senior people stay involved after the sale, and the firm's reputation concentrates in a referenceable handful of clients who can actually be called. Rates typically run below the enterprise tier's, because the overhead structure is lighter, while the seniority actually deployed on a mid-size engagement is often higher, because the firm's best people are not reserved for hundred-person accounts.
The segment's risks are the mirror of its strengths, and honest vendors name them. Concentration: a small firm losing two large clients in a quarter feels it, so buyers should ask about client concentration and financial runway. Depth ceilings: a 200-person firm cannot staff a 150-person program, and should say so rather than try. Key-person dependency: the senior architect who impressed you in the sales process is a real single point of failure, so contractual named-person stability, the same clause the offshore guide recommends, protects both sides. These are manageable risks, and the management is exactly the diligence the evaluation guide covers.
Selection inside the segment is where domain evidence does the work. A specialist's claim to fintech or healthcare depth is checkable: case studies with named or referenceable clients, engineers who can discuss the domain's hard problems unprompted, in an interview, not a deck, and opinions about the domain's regulatory and integration realities that could only come from having shipped in it. Firms with real specialization pass this test conversationally; generalists dressed as specialists reach for the marketing language, and the difference is audible within an hour.
The segments against buyer profiles
| Your engagement | Best-fit segment | Why |
|---|---|---|
| Hundred-person, multi-year enterprise program | Large domestic firms | Staffing depth, certified process, procurement-grade structures |
| Five-to-thirty-person product team, ongoing | Mid-size specialists | Senior attention, direct communication, domain depth per dollar |
| Short fixed-scope project | Mid-size specialists or agencies | Project economics; the team model's ramp never repays |
| Own entity, direct hiring | You compete with the giants | Price offers against global sites and product companies, not services salaries |
The fit logic in one table: match the engagement's shape to the segment's economics before comparing any two vendors.
Why the published top-ten lists mislead
The listicles that dominate this search deserve their own section because they actively distort buying decisions. The mechanics are simple: most ranking directories monetize through vendor subscriptions, sponsored placements and lead fees, and most listicles on agency blogs rank the publishing agency first. The result is not that listed firms are bad, many are fine, but that presence and position on the list measure marketing spend, not delivery quality, and a buyer treating rank as evaluation has outsourced diligence to an advertising channel.
Review platforms are the more defensible version of the genre and still need reading skill. Verified-review sites carry real client interviews, and the interviews contain genuine signal, but the aggregate scores compress everything to a number where nearly every vendor holds 4.8 stars, review volume tracks how systematically a firm harvests reviews, and unhappy clients of any vendor rarely write reviews at all. The useful method is to ignore scores, read the negative and middling reviews first, and mine the review text for the specifics that match your engagement: team size, domain, engagement length, what went wrong.
The deeper problem with any list is that top has no denominator. Top for whom, at what size, in which domain, under which model? The segment analysis this page runs exists precisely because the question has no single answer: FPT is genuinely a top company for a global enterprise program and genuinely the wrong choice for a seed-stage startup's first team, and both facts fit in one sentence. Any list that ranks them against a 200-person fintech specialist on one axis is compressing away the only information that matters.
What replaces the list is a process, and it is shorter than it sounds: define the work's shape, size, duration, domain, model; pick the segment whose economics fit; assemble five to eight candidates from checkable sources, referenceable case studies, alumni networks, peer recommendations, the review text mined as above; and run the partner-level diligence, attrition, tenure, ownership culture, design-review pushback, that the partner evaluation guide details. Two weeks of that process outperforms any ranking ever published.
Using the directories without being used
Do this
- Treat lists as a phone bookA discovery source for names you then evaluate independently. Presence means marketing budget, nothing more.
- Read negative reviews firstThe middling and unhappy reviews carry the signal; the five-star wall is harvested.
- Mine reviews for your shapeTeam size, domain, duration, what went wrong. A great review of a forty-person program says little about your six-person team.
- Verify off-platformReferenceable clients you can call, alumni you can check, code or case detail an engineer can interrogate.
Not this
- Shortlist by rank positionRank measures directory spend. You are reading an ad and calling it research.
- Trust aggregate scoresWhen every vendor scores 4.8, the metric carries no information by construction.
- Compare across segments on priceAn enterprise firm's rate buys different machinery than a specialist's. The comparison needs the segment context.
- Skip the diligence for a brandBig names fail small engagements structurally, not accidentally. Size fit beats brand comfort.
Building a real shortlist: the process
Start from the work, not the market. Write one paragraph: what is being built, how many people it needs, for how long, in what domain, under which engagement model, team extension, project delivery, or staff augmentation, and which client geography's working hours it must meet. That paragraph makes the segment choice nearly automatic through the fit table above, and it becomes the brief candidates respond to, which itself is a filter: vendors who answer the brief specifically are telling you how they will communicate for the next two years.
Source candidates from channels where incentives are aligned. Peer recommendations from companies with engagements shaped like yours are the strongest channel. Referenceable case studies, where the client is named and callable, are next. Alumni networks, engineers or founders who have worked with or inside Vietnamese firms, carry dense, honest signal. Review text mined for engagements like yours, as the previous section described, is usable. Cold directory rank is the weakest channel and should contribute names only, never ordering.
Run the evaluation as an engineering exercise, not a procurement theater. A structured call where your engineers interview theirs, on a real problem from your domain, tells you more than any deck: listen for specific past examples versus generalities, for questions they ask you, and for disagreement, the design-review pushback test from the offshore guide. Follow with references you choose from their client list rather than the two they offer, and ask every reference the same three questions: what went wrong, how was it handled, would you start again with them tomorrow.
Then pilot before you commit, if the engagement size justifies it. A four-to-eight-week paid pilot on a real, bounded piece of work, with the actual proposed team, converts every claim into observable behavior: communication rhythm, code quality under your review bar, how estimates track reality, how the first disagreement lands. Pilots cost a fraction of a mis-signed annual contract, and good vendors welcome them, because the same pilot is how they evaluate you as a client, and the ones who resist a paid pilot on principle are answering a question you should notice.
From market to shortlist in five moves
-
Write the work paragraphHalf a day
Scope, size, duration, domain, model, working hours. The segment choice falls out of it, and it becomes the brief.
-
Source from aligned channelsA week
Peers with similar engagements, callable references, alumni networks, mined review text. Directories contribute names only.
-
Interview engineer to engineerOne call each
A real domain problem, your engineers versus theirs. Listen for specifics, questions back, and honest disagreement.
-
Call references you chooseThree calls
From their client list, not their shortlist. What went wrong, how was it handled, would you start again.
-
Pilot the actual teamBefore the big contract
Four to eight weeks, paid, real bounded work, the named people. Claims become observable behavior.
Frequently asked questions
What are the biggest software companies in Vietnam?
FPT is the flagship: its global IT services revenue passed one billion dollars annually in 2023 and it employs tens of thousands of engineers worldwide. Behind it sit CMC, Viettel's software arms, and a tier of firms with thousands of engineers including TMA Solutions, KMS Technology, NashTech and Rikkeisoft. Among product companies, VNG (Zalo), MoMo and Sky Mavis anchor the segment, and Samsung's Hanoi R&D center is the largest multinational engineering site.
Are the published top-ten lists of Vietnamese vendors reliable?
Treat them as advertising. Most ranking directories monetize through vendor subscriptions, sponsored placements and lead fees, so presence and rank measure marketing spend rather than delivery quality. Review platforms carry real signal, but it lives in the review text, especially the negative and middling reviews, not in aggregate scores where nearly every vendor holds 4.8 stars. Use directories to discover names, then evaluate independently.
What kind of Vietnamese firm fits a foreign product company?
Usually a mid-size specialist: 50 to 1,000 engineers, serving foreign clients directly, with a recognizable domain or stack center of gravity. At that scale your team is a meaningful account, senior people stay involved after the sale, communication is direct rather than account-managed, and rates run below enterprise-tier pricing. The enterprise tier fits large multi-year programs; placing a small product team there buys process designed for something else.
What does Samsung's R&D presence in Vietnam signal?
Talent validation at the highest standard of evidence. Multinationals audit engineering markets harder than any procurement process before committing R&D sites, and Samsung's Hanoi center, among its largest outside Korea, plus sites from LG, Bosch, Panasonic and others, is that audit's published result. The sites also train engineers to global practices and push the whole market's salary curve upward, both of which buyers feel downstream.
How should a buyer actually shortlist Vietnamese vendors?
Work backward from the engagement: write one paragraph covering size, duration, domain, model and working hours, pick the segment whose economics fit, then source five to eight candidates from aligned channels, peer recommendations, callable references, alumni networks and mined review text. Evaluate engineer to engineer on a real domain problem, call references you choose from their client list, and run a paid pilot with the named team before the big contract.
Do Vietnamese firms build products, or just execute specifications?
The product segment answers this directly: VNG operates Zalo at a hundred-million-user scale, MoMo runs one of Southeast Asia's largest payment platforms, Sky Mavis built a globally significant blockchain game, and the games industry ships worldwide hits. Services firms hire from the same talent pool, and the practical test for any vendor is whether the proposed team includes engineers with product-company or global-site backgrounds, which is checkable.
The top-ten lists of Vietnamese software companies are mostly pay-to-play advertising, and the real landscape is four segments serving different buyers. Before shortlisting anyone, see the buyer's map of Vietnam's software industry, from Samsung's R&D site and FPT to the specialist tier where most foreign buyers actually fit.
