Every time NVIDIA ships a Blackwell GPU or SK Hynix stacks a new HBM4 module, somewhere in that process an Onto Innovation inspection system has scanned the die, measured the bump height, and checked that the 3D stack will actually work. Most investors follow NVIDIA. Almost nobody follows ONTO. That gap is the opportunity.

Onto Innovation is a semiconductor process control company — it makes the systems that check chips for defects and measure critical dimensions during manufacturing. It is not a household name. But inside every leading-edge fab and advanced packaging line in the world, Onto's tools are running 24 hours a day. And as AI chips get more complex, the value of those tools goes up with every new node generation.

The Numbers That Matter Right Now

Q1 2026 Revenue
$291.9M
Beat guidance of $275–$285M
Q2 2026 Guidance
$320–330M
+8% above prior outlook
Non-GAAP Gross Margin
55.7%
Q1 2026
HBM Contract
$240M+
Volume purchase agreement

The Q1 beat was not a fluke. Onto came into 2026 with $1.005 billion in full-year 2025 revenue — a modest 2% growth year — and immediately accelerated. Management raised Q2 guidance 8% above what they had communicated just weeks earlier, which in the semiconductor equipment business is a significant signal. These companies do not raise guidance casually.

Why the Dragonfly G5 Is a Much Bigger Deal Than It Sounds

In April 2026, Onto announced that its Dragonfly G5 inspection system had passed TSMC's New Tool Selection Committee qualification process for 2.5D advanced packaging. That single sentence deserves unpacking.

TSMC's advanced packaging — CoWoS, SoIC, and related stacking technologies — is the platform that makes NVIDIA's AI accelerators physically possible. You cannot sell an H100 or a B200 without advanced packaging. TSMC's qualification process is one of the most demanding in the industry. When Onto passed it, the company effectively got a license to print money on every 2.5D AI chip TSMC produces going forward.

On top of that, a leading HBM manufacturer selected the Dragonfly G5 for 2D inspection for the HBM4 ramp, with double-digit orders already committed and shipments beginning in Q2 2026. Onto also secured a separate volume purchase agreement for HBM inspection — 2D plus 3D metrology — worth over $240 million, with more than $60 million going to 3D systems alone.

Key insight: Dragonfly platform demand is expected to grow more than 50% in 2026 versus 2025. This is not incremental growth — it is an inflection driven by the transition to HBM4 and CoWoS-L packaging architectures, both of which require dramatically more inspection steps than prior generations.

Advanced Packaging Is the New Battleground — And Onto Owns the Inspection Layer

Here is why advanced packaging inspection is so valuable: when chipmakers were just shrinking transistors on a flat die, inspection was important but manageable. Now, with 3D stacking, you are bonding multiple dies on top of each other with thousands of microscopic copper pillars (called microbumps or hybrid bonds). If even one of those connections is misaligned or defective, the entire multi-thousand-dollar chip is dead.

The inspection tolerance required for HBM4 and CoWoS-L packaging is measured in nanometers. No human eye can check this. No conventional camera can resolve it. You need Onto's specialized optical and laser-based systems, and you need them at every stage of the process — before bonding, during bonding, and after bonding.

TSMC has publicly guided for advanced packaging demand to grow roughly 30% annually as AI accelerator complexity increases. Every percentage point of that growth translates directly to more Onto inspection tool purchases. The company is, as one analyst put it, the advanced packaging chokepoint — you simply cannot ramp AI chip production without it.

The Semilab Acquisition: Buying the Next Growth Layer

In 2025, Onto acquired Semilab, a metrology specialist with particular depth in compound semiconductors, silicon carbide, and gate dielectric measurement. Management expects Semilab to contribute approximately $120 million in FY2026 revenue. That is not a small number for a company that did $1 billion in total revenue last year.

More importantly, Semilab expands Onto's addressable market into silicon carbide — the material used in EV power electronics and high-frequency chips — and into advanced gate dielectric measurement for the next generation of transistor architectures (Gate-All-Around, nanosheet FETs). These are not niche applications; they are the next decade of semiconductor manufacturing.

What Could Go Wrong

Onto is not without risks. The semiconductor equipment industry is notoriously cyclical, and while the AI buildout appears durable, any slowdown in hyperscaler capital spending would ripple directly to tool orders. The company is also in an intensifying competition with KLA Corporation, which has deeper pockets and a broader product portfolio across defect inspection. Onto wins on speed and specialization in advanced packaging, but KLA is not standing still.

Customer concentration is another concern. A meaningful portion of Onto's revenue comes from a small number of leading-edge customers — TSMC, SK Hynix, Samsung, and Micron. If any one of those customers pushes out a ramp or shifts tool suppliers, it shows up quickly in Onto's quarterly results.

The Semilab integration carries execution risk, as all acquisitions do. And with volume Dragonfly G5 shipments starting mid-2026, there is a back-half revenue ramp baked into market expectations. If shipments slip a quarter, the stock will feel it.

The Bigger Picture

The AI infrastructure buildout is most visibly about GPUs, data centers, and power. But underneath all of that is a manufacturing precision layer — and Onto Innovation is one of the companies that makes it physically possible to produce chips at the tolerances AI demands. The Dragonfly G5 TSMC qualification, the HBM4 wins, and the raised guidance are not isolated events. They are the beginning of a sustained cycle where AI complexity forces fabs to buy more inspection equipment per wafer start, not fewer.

Onto Innovation sits at that intersection. It is not the flashiest name in the AI trade. But in a world where yield at advanced packaging nodes is the difference between billions in profit and billions in scrap, the company that ensures every chip passes inspection before it ships is indispensable. That kind of indispensability tends to show up in earnings over time.

This article is for informational purposes only and does not constitute investment advice. Investing in securities involves risk, including possible loss of principal. Always conduct your own research and consult a qualified financial advisor before making investment decisions.