DEV Community

SleepyQuant
SleepyQuant

Posted on • Originally published at sleepyquant.rest

FPT Corporation and the AI Consulting Margin Compression: Why Vietnam's Biggest Tech Firm Lost a Third of Its Market Cap

#ai #machinelearning #finance #programming

FPT Corporation and the AI Consulting Margin Compression: Why Vietnam's Biggest Tech Firm Lost a Third of Its Market Cap

An IT Giant Most Western Investors Have Never Heard Of

FPT Corporation, Vietnam's largest IT services firm, is down ~33.8% from its 52-week high. This drawdown mirrors a broader sector-wide slump: TCS fell 21.4%, Wipro dropped 23.1%, and Infosys declined roughly 16% over the same window. The market appears to be repricing the entire labor-arbitrage consulting model at once, not punishing FPT in isolation.

Here's what makes it interesting: in 9M2025, FPT still grew revenue +10.3% YoY (VND 49,887 billion ≈ $1.96B USD) and pre-tax profit +17.6% YoY (VND 9,540 billion ≈ $374M USD). The fundamentals didn't crash. The expectations did.

I went down this rabbit hole after watching Mèo Giải Thích's Vietnamese-language deep dive on FPT (388k+ views). What follows is a case study in AI consulting margin compression — one of the cleanest sector-wide pricing events I've seen in IT services in the past year. Below: what FPT actually does, the AI catalyst that hit the entire sector at once, and the counter-case the market isn't pricing in.

Bar chart: FPT down 33.8% from 52-week peak (worst), Wipro -23.1%, FPT 1Y -22.2%, TCS -21.4%, Infosys -16.5%. AI margin compression hit the entire labor-arbitrage IT consulting sector simultaneously.

What FPT Actually Does

From banana flour machines to Vietnam's largest IT firm

The founding story is almost too literal to be real: in 1988, the acronym FPT stood for "Food Processing Technology." Early FPT was drying cigarettes and installing air conditioners. Then came the pivot in 1990 — a $1 million computer contract with the Soviet Academy of Sciences changed everything. Within roughly a decade, FPT had become Vietnam's dominant IT firm. Understanding their current engine, though, requires looking at three distinct pillars rather than the single "IT" label.

Three pillars: Technology, Telecom, Education

According to the official 9M2025 earnings report (~$1.96B USD in nine-month revenue), Technology remains the undisputed core: about 62% of group revenue and 45% of group pre-tax profit. Telecom follows as a steady cash cow, contributing 29% of revenue (≈$539M USD) with surprising margin expansion — pre-tax profit grew +21% despite limited market-size headroom. Education rounds out the trio at just 9%; historically high-margin, but recent stagnation hints at real competitive pressure (more on that next).

Why Japan is FPT's biggest customer

What fascinates me most about FPT's Tech segment is where the money actually lives: overseas markets capture roughly 80–90% of that division's inbound revenue. Japan sits firmly as #1, followed by the US and APAC. Why? Because demographic collapse there has created an IT labor shortage so severe that Japanese planners are now recruiting half a million Indian tech workers to fill the gap. FPT's labor-cost advantage is the bridge Vietnamese firms have been crossing for years. In 2024, FPT also opened two AI factories — one in Vietnam, one in Japan — but they're still too small to materially move group numbers.

The Margin Story Hidden in Education

Education accounts for just 9% of FPT's total revenue, yet it has historically been a cash cow with pre-tax margins hovering between 40-50%, according to Mèo Giải Thích. This profitability stems from a vertically integrated talent pipeline: FPT operates schools ranging from K-12 through university, and many graduates join the company directly. By internalizing recruitment, FPT drastically reduces external hiring friction and retraining costs while ensuring new hires are already aligned with its specific technical culture and operational standards. It is an elegant self-sustaining loop where education fuels technology growth without depending on volatile external labor markets.

However, the official 9M2025 earnings data reveals a sharp divergence from that high-margin narrative. Education revenue grew only +1.0% YoY to VND 5,195 billion (≈$204M USD). This stagnation suggests headwinds are biting harder than anticipated. Vietnam's K-12 fee waiver in public schools has eroded the addressable market for private tuition, with families increasingly opting for free state alternatives over premium rates at FPT institutions.

Why The Market Lost Faith — The P/E Compression

From 30x to 15x in eighteen months

I first understood valuation through a coffee-shop analogy from Mèo Giải Thích's video: if a shop earns $1 million a year but sells for $20 million, the Price-to-Earnings ratio is 20. Buyers are paying twenty years of current profits upfront for the future growth they expect.

FPT's stock chart tells the same story in real time. P/E peaked around 30x when optimism was highest, normalized to roughly 19x over recent quarters, and now sits near 15x. The compression signals that investors have drastically lowered their growth assumptions.

How FPT compares to Indian IT consulting peers

Here's the part that should make any FPT bull pause: the Indian IT consulting comparables aren't trading much higher. As of April 2026, TCS sits around ~19x trailing P/E, Infosys around ~18x, Wipro around ~16x. FPT at ~15x is trading at a discount to all three. Sector-wide compression explains most of the move, but FPT carries an additional discount on top — the market is pricing in either smaller scale, less diversified revenue base, or company-specific execution risk that its global peers don't have.

Bar chart: FPT trades at ~15x trailing P/E, Wipro 16x, Infosys 18x, TCS 19x. FPT trades at a discount to all three Indian IT consulting peers as of April 2026.

What the official 9M2025 numbers actually show

The official 9M2025 data backs the deceleration. Tech segment revenue grew only +10.7% YoY against the segment's 24% historical CAGR (2018-2024). Total group revenue reached VND 49,887 billion (≈$1.96B USD) — still positive, but well off the trajectory the old multiple priced in. The gap between former hype and current reality is why the multiple collapsed from 30x toward 15x. But P/E compression doesn't happen in a vacuum — there was a specific catalyst.

Why The Market Lost Faith — The AI Catalyst Behind the Margin Compression

February 23: The Anthropic shot heard around IT services

The catalyst for the sector's re-rating arrived on February 23, 2026, when Anthropic published "How AI helps break the cost barrier to COBOL modernization". They claimed Claude Code could map dependencies across thousands of lines of legacy code, document workflows, and identify risks that "would take human analysts months to surface." This was not an abstract tech update — it was a direct shot at the consulting layer where firms charge premium hourly rates for human-led modernization work, exactly FPT's core moat in digital transformation and system integration.

IBM down 13.2% in a single day, FPT followed

The market reacted immediately: IBM stock fell 13.2% that same day. The pricing signal suggested investors were rapidly discounting future labor-arbitrage margins across global IT services providers. FPT's own decline accelerated after this date — the timing is suggestive rather than coincidental within the broader -16% to -23% sector drawdown seen across TCS, Infosys, and Wipro. An insider sale by board member Bùi Quang Ngọc near the peak (timing flagged by the Mèo Giải Thích video) drew attention, but the source video itself cautioned against over-reading: founders Trương Gia Bình and Đỗ Cao Bảo did not sell, and a single insider transaction without volume context is closer to noise than signal. (For a tangentially-related build-in-public take on how a single missing argument in production code can compound into outsized loss, see my recent post on a one-line trading bug.)

The Counter-Case Worth Hearing

AI doesn't only delete consulting — it creates new categories

The bear case assumes AI simply deletes consulting hours, but it also creates entirely new categories. Companies still need someone to deploy these models into actual operations, integrate complex stacks with legacy systems, and train staff on the new workflows. FPT has explicitly pivoted in response, declaring an "AI-first" strategic direction. They are building infrastructure rather than relying on manual code migration alone. Their flagship vehicle is the FPT AI Factory, positioned as a "one-stop shop" for AI and Cloud services. At CES 2026, FPT showcased AI-first innovations across industries from automotive to semiconductor design.

What FPT's own numbers say about the pivot

Per FPT's own reporting, their AI and Data Analytics service lines grew +41% YoY — tangible demand, though importantly off a small base; AI services are still single-digit percent of group revenue, not yet large enough to offset the deceleration in legacy Tech consulting. Chairman Trương Gia Bình has publicly emphasized future bets on Quantum Computing, Cybersecurity, UAVs, and Railway Tech, all underpinned by core AI capabilities. Two AI factories are operational — one in Vietnam, one opened in Japan in 2024 — aimed directly at the labor shortages and digital transformation demand that have driven FPT's overseas growth for years. At a ~15x P/E, the market is pricing low odds that this pivot scales fast enough. That is where the optionality sits.

What I'm Watching, From Outside

I am trying to understand FPT as a business, not as a ticker. The recent drawdown is stark, but the real story lies in three leading indicators that reveal whether the company can pivot from legacy arbitrage to AI-driven value:

  • New Contract Value (NCV) — the leading indicator for future Technology segment revenue. If NCV stagnates while signed revenue keeps growing through backlog consumption, that's demand friction showing up before it hits the top line.
  • Tech segment pre-tax margin trend — the canary for AI pricing pressure. As AI tools compress billable hours per project (the same dynamic behind IBM's 13% drop), it shows up here long before it shows up in total sales volume.
  • AI Factory contribution to group revenue — the strategic execution check. If the two factories (Vietnam + Japan) can move from single-digit % to materially mid-single-digit % over the next 4-8 quarters, the bull pivot is landing.

None of this is a recommendation. I'm watching because the case is interesting, not because I have an edge. The cost economics are also why I keep coming back to the Apple Silicon angle on local AI — the same dynamic that compresses FPT's consulting margins is what makes running a 35B model on a laptop suddenly viable. Credit again to Mèo Giải Thích for doing the heavy synthesis on the Vietnamese-language side; this post is my attempt to put that story into English with the official 9M2025 earnings numbers cross-checked.

The Lesson Beyond FPT

Pulling back from the company-specific drama reveals a sector-wide structural shift. Tata Consultancy Services is down 21.4%, Infosys 16.5%, Wipro 23.1%, FPT 22.2% over the same window — peak-to-trough on FPT is closer to 33.8%. Four major labor-arbitrage IT consulting firms across two continents, all repricing in the same direction at the same time.

This is not a Vietnam story or even an FPT story. It is the entire "humans do the consulting work" business model getting re-rated by AI. The survivors will pivot fast from "we sell hours" to "we sell the AI that does the hours". The casualties will stay too long in the now-commoditizing layer.

The IBM chart on Feb 23 and the FPT chart in the weeks after are saying exactly the same thing.

Sources

This analysis was anchored on a Vietnamese-language YouTube video by Mèo Giải Thích (Explaining Cat), a Vietnamese economics explainer channel — the primary narrative spine. The hard numbers and corroborating data points came from the following public sources:

This is independent analysis grounded on publicly available sources. Not financial advice. Numbers stated are as of the source date noted; equity prices move continuously and any specific level cited may be stale by the time you read this. The author holds no position in FPT Corporation, Tata Consultancy Services, Infosys, Wipro, or IBM at the time of writing. Mèo Giải Thích is credited as the anchor source and was not consulted for this article.

If this was useful, I write weekly at sleepyquant.rest. One email a week, real numbers, no signals. Subscribe — come along to see me fall or thrive, whichever comes first.

Top comments (0)