GIGABYTE turns 40 this week. The press releases highlight four decades of motherboards and innovation. Investors should look at something else: the company just reported all-time record earnings on an AI server tailwind, and the stock's valuation has run ahead of what that growth can safely support.
I'm not upgrading. The stock is a Hold - the business is better than most people think, but the price already reflects a flawless execution scenario.
What changed
GIGABYTE's Q1 FY2026 results, released in mid-May, show a company caught in the best kind of growth cycle. Revenue jumped 60% year over year to NT$105 billion (roughly $3.3 billion). AI server revenue surged 110% year over year. More importantly, AI servers now account for approximately 71.5% of total sales and roughly 90% of operating profit - meaning the margin mix has fundamentally shifted away from the low-margin consumer motherboard and graphics card business that defined the company for most of its history.
Gross margin climbed to 12.01%, up 1.75 percentage points from the prior quarter. Profit attributable to shareholders hit NT$5.27 billion, with EPS of NT$7.86 - both all-time highs. Full-year 2025 revenue set a company record at NT$336.9 billion.
These numbers aren't cosmetic. The shift from consumer hardware to AI infrastructure is real and it's accelerating.
The margin picture is improving - for now
The 12% gross margin is still thin by software standards, but for a hardware integrator assembling expensive GPU servers, it's competitive and rising. That matters because the alternative scenario in this space is brutal margin compression. Supermicro, a direct peer in AI servers, saw gross margins collapse to 6.3% in early 2026 despite $12.7 billion in quarterly revenue - crushed by hyperscaler pricing pressure from customers who dictate terms. GIGABYTE's margin trajectory is moving in the opposite direction, which suggests better product mix discipline or less hyperscaler concentration risk - or simply that the market is rewarding GIGABYTE for what it already accomplished.
The company is also building a US server assembly line to fulfill a large order from CoreWeave. That's a hedge against tariff exposure and a signal that enterprise AI cloud builders still trust GIGABYTE's engineering.
Where the stock gets expensive
Here's the problem. GIGABYTE shares trade around NT$322 on the Taiwan exchange, with a market cap near NT$225 billion (about $6.9 billion in USD). The trailing P/E sits at roughly 19.5x - and that's 38% above the stock's own 10-year median P/E of about 13.4x.
In other words, the market has re-rated this stock upward by a wide margin. A hardware assembler with 12% gross margins trading nearly 20 times earnings is paying a multiple that assumes sustained acceleration. It's not a bubble, but it's not a bargain either. The stock needs continued AI server growth and stable or expanding margins just to justify where it sits now.
Ten analysts covering the Taiwan-listed shares have a consensus Buy rating (7 buys, 3 holds, zero sells). That's the kind of unanimity that usually precedes periods of flat or disappointing returns. When everyone already likes the stock, the risk/reward shifts.
The risks nobody in the press release mentions
- Customer concentration. AI server demand is driven by a small pool of hyperscalers and cloud builders. If NVIDIA's GPU supply chain reshuffles preferred vendors, or if one major customer delays deployments, revenue swings hard.
- Margin durability. The 110% AI server growth rate is a fast decelerator. You can't compound at that pace forever. Once growth normalizes to 30-40%, the P/E expansion that fueled the stock rally becomes unjustified unless margins widen further.
- Consumer segment drag. The consumer hardware business (motherboards, GPUs, AORUS gaming gear) continues to decline. Revenue fell in the consumer segment last quarter. The brand that made GIGABYTE famous is a smaller part of the business every year, and it's shrinking.
- No US listing convenience. GIGABYTE trades on the Taiwan exchange. The old NASDAQ ADR ticker (GIG) is no longer the company - it was a different entity. US investors need a Taiwan broker or an OTC vehicle, which limits liquidity and institutional access.
The catalyst clock
Q2 FY2026 earnings are expected around mid-August 2026, based on the company's prior reporting pattern. That's the next proof point. The question won't be whether AI server revenue grows - it will be whether margins hold or begin compressing as more competitors enter the space and hyperscalers push harder on pricing. The US CoreWeave production line should also be operational by then, and management commentary on its economics will matter.
Investor takeaway
GIGABYTE is not the brand-celebration story the PR machine is selling. It's a Taiwan hardware assembler that successfully pivoted into AI servers and is reaping the rewards - higher revenue, better margins, record EPS. The problem is that the stock price has already done its work. At 19.5x earnings, well above its decade-long average, the multiple assumes the AI tailwind stays intact and margins don't slip.
That's a Hold, not a Buy. The business deserves respect, but the entry price doesn't offer enough cushion for execution risk. I'd wait for a pullback toward the 14-16x P/E range before initiating a position - a zone that gives room for error if AI server growth decelerates faster than expected. Conversely, if Q2 shows margin compression or a guidance cut, the current multiple would prove too rich, and a downgrade would follow.
Monitor: Q2 FY2026 gross margin trend, AI server revenue growth rate deceleration, and any changes in the NVIDIA supply chain vendor hierarchy.