Why We Build With Yanmar Engines and Bosch Hydraulics Instead of Our Own

We use proven, world-class diesel engines and hydraulic systems so we can focus on building AI that actually changes how work gets done on a jobsite. Here is why that matters for you.

---

The Question We Get Asked Most

When people first learn about Grizzly, they expect us to have designed everything from scratch. Custom engine. Proprietary hydraulics. Our own transmission. That is the playbook most hardware startups follow: control every layer of the stack, own every component, differentiate on everything at once.

We went the other direction. Our machines run Yanmar and Kubota diesel engines. Our hydraulic systems are Bosch Rexroth. And we are not shy about it.

This is not a compromise. It is a deliberate engineering decision, and we think it is one of the most important ones we have made.

First Principles: Build What Doesn't Exist Yet

When we started Grizzly, we asked a simple question: where can we build something that doesn't exist yet, rather than rebuilding something that already works?

Diesel engines are not that place. Yanmar has been building compact diesel engines since 1933. They have nearly a century of iteration on combustion chamber geometry, fuel injection timing, thermal management, and emissions compliance. Their engines power millions of machines worldwide. The failure modes are catalogued. The maintenance procedures are documented. The parts supply chain spans every continent.

Hydraulic systems are not that place either. Bosch Rexroth has been engineering hydraulic components since the 1950s. Their axial piston pumps, proportional valves, and load-sensing circuits represent decades of metallurgical research, fluid dynamics optimization, and field validation across every heavy industry on the planet.

Could we build our own engine? Probably. Could we design our own hydraulic system? Sure. But "could" is not the same as "should."

The place where genuinely new value gets created is in the intelligence layer. Not autonomy that does one thing — auto-grade a pad, dig a straight trench — but intelligence that handles a real day of work. An excavator that adapts to what the job demands, the way a good operator does, across every type of dirt moving task a site throws at it. That is where construction equipment has barely scratched the surface, and that is where we pour our engineering resources.

The Tesla Parallel

This is not a new idea. Tesla did not invent tires. They did not design their own brake calipers or windshield glass or seat foam. They used proven, world-class components from established suppliers for everything that was not their core breakthrough.

Their breakthrough was the battery system, the power electronics, the software, and eventually autonomous driving. That is where they focused, and the results speak for themselves.

Construction is not consumer tech — the stakes are higher and the conditions are harsher. But the principle of focusing engineering resources where you create genuine differentiation applies.

Our breakthrough is the AI that makes an excavator useful across an entire shift — not just for one scripted task. Sensor fusion that builds a real-time 3D understanding of the terrain. Voice control that lets you direct your machine the way you would talk to your best operator. Autonomous operation that adapts to what the job demands, whether that is grading, trenching, loading, or shaping.

We would rather pair a proven Yanmar engine with intelligence that keeps getting smarter than split our team between reinventing the diesel engine and building software that only does one thing well.

What This Means for You on the Jobsite

Here is where the decision to use proven components stops being an engineering philosophy and starts being a direct benefit to the people running our machines.

Serviceability You Already Know

Your crews have worked on Yanmar engines before. Your local dealer stocks Yanmar parts. Your mechanics have the manuals, the training, and the muscle memory. When a Grizzly machine needs an oil change or a fuel filter, it is the same procedure they have done hundreds of times.

This is not a small thing. Proprietary powertrains create a service dependency that can strand a machine on a jobsite waiting for a specialist. We have watched that happen to contractors running other equipment, and we refuse to put our customers in that position.

Known Failure Modes

When you run components with decades of field history, you get something invaluable: predictability. The failure modes of a Yanmar 4TNV88C are well-documented across millions of operating hours. Your mechanics know what to watch for. Your maintenance schedules are proven. Your parts inventory decisions are backed by real data, not guesswork.

Compare that to a first-generation proprietary engine with a few thousand hours of testing. You become the beta tester. Your jobsite becomes the proving ground. That is a risk we are not willing to pass on to you.

Supply Chain Resilience

The last several years have taught every manufacturer a hard lesson about supply chains. When you depend on a single source for a custom component, one disruption can halt your entire production line and leave your customers without machines or parts.

Yanmar and Bosch Rexroth have global manufacturing footprints with multiple production facilities, established raw material supply agreements, and decades of logistics infrastructure. Their components are produced at scale for dozens of OEMs, which means production continuity even when individual markets face disruption.

Our machines inherit that resilience. A Grizzly customer is never waiting because a single-source proprietary component is backordered.

Where We Do Innovate

The engineering resources we do not spend on powertrains go somewhere specific. Here is what we build and, more importantly, why each piece requires dedicated focus.

Sensor fusion and perception. A single sensor type is not enough on a construction site. LiDAR gives you geometry but struggles with dust and rain. Cameras give you context but lose depth accuracy at distance. Radar punches through weather but lacks resolution. The hard engineering problem is not mounting sensors on a machine — it is fusing their data streams in real-time so the machine builds one coherent, centimeter-accurate model of the jobsite, even when individual sensors disagree or degrade. That fusion pipeline is why we need a dedicated perception team, and it is something no off-the-shelf integration can solve for the conditions heavy equipment actually operates in.

Autonomous operation. The question we keep asking is not "can we automate one task?" but "can we handle a real day?" Variable soil density, cut-to-fill transitions, terrain that changes with every pass, switching between grading and trenching and loading as the job requires. Our planning and control systems make thousands of micro-decisions per second. The bulk of our software engineering lives here because the gap between a demo and a machine a contractor trusts on a real job is enormous.

Natural language interface. We invested in voice control not because it is a flashy feature, but because it solves a real architectural problem: how do you let a non-specialist direct an autonomous machine without weeks of training on proprietary software? "Grade this pad to minus two percent slope" is a valid command. The machine understands it, plans the work, and asks clarifying questions if the instruction is ambiguous. This matters because the industry is short on experienced operators, and the interface is the bottleneck.

Fleet coordination. When multiple machines work the same site, you either coordinate them manually — radios, hand signals, a superintendent watching everything — or you build a system where the machines coordinate themselves. We chose the latter. Shared terrain data, conflict avoidance, and collective path optimization are not features we added for marketing. They are the logical consequence of deciding that intelligence should live at the fleet level, not just the machine level.

Continuous learning. Every hour of operation generates training data that feeds back into our models. This is an architectural decision, not just a product feature. It means customers who bought a machine six months ago are running better software today than the day it shipped, without changing a single mechanical component. The powertrain stays the same. The intelligence compounds.

None of that exists inside a Yanmar engine or a Bosch hydraulic valve. All of it exists because we chose to focus our engineering talent where the hard, unsolved problems are.

The Honest Tradeoff

We should be transparent about what we give up with this approach. We do not control every variable in the powertrain. If Yanmar changes a specification, we adapt. If Bosch updates a valve design, we integrate it. We are downstream of their roadmaps on those components.

We accept that tradeoff gladly. The alternative is splitting our engineering team between building AI that does not exist yet and rebuilding engines that already work. That math does not add up for us, and it does not add up for our customers.

What Being Pro-Customer Actually Looks Like

We built the company where your mechanic can swap a fuel injector without calling us. Where your parts supplier already stocks what you need. Where your service infrastructure works on day one because the mechanical components are familiar. And where the thing that is genuinely new — the AI that makes your crew faster and your bids more competitive — gets every dollar of engineering investment we can give it.

That is first-principles thinking applied to building a company, not just a machine. Spend resources where you create outsized value. Use the best existing solutions for everything else. Be honest with your customers about what you build and why.

We build with Yanmar engines and Bosch hydraulics because they have decades of proven performance in exactly the conditions our machines operate in. We build our own AI because nobody else is building what we are building. And we tell you all of this because we think you deserve to know exactly what you are buying and why it was built that way.

---

For engineering inquiries or to discuss our component strategy in more detail, reach out at research@usegrizzly.com.