Last Mile Semiconductor is building the chip that makes that happen, and it is why we are so excited to be backing Christoph, Mario, and the team.
The compromise nobody chose
For twenty years, connecting a device wirelessly has meant picking your poison. Long range or high throughput. Battery life or bandwidth. Free spectrum or reliability. Every protocol on the market today is a compromise, and every connected product is shaped by whichever limitation the engineers decided they could live with.
Bluetooth falls apart beyond ten meters. WiFi drains batteries in hours and chokes on its own interference in dense deployments. Cellular gives you range and throughput but locks you into SIM cards, base stations, monthly fees, and someone else's network. Low-power wide-area options like LoRa reach kilometers but can barely carry a temperature reading.
The billions of "connected devices" the industry has been promising for a decade? Most of them are not actually connected. They sit behind proprietary gateways or whisper a data packet once an hour because the radio they are stuck with cannot do anything more.
A standard that changes the maths
Most people over thirty have owned a DECT cordless phone. The frequency band those phones used, 1.9 GHz, has been sitting in license-free spectrum in most of the world for decades, quietly doing nothing interesting.
In 2021, the European Telecommunications Standards Institute (ETSI) took that spectrum and built something entirely new on it: DECT NR+, a non-cellular wireless standard in the same 5G family as the networks in your phone, but designed from scratch for machines instead of people. No operators. No SIM cards. No infrastructure to deploy. Devices form their own mesh networks, finding each other and rerouting around failures automatically.
The performance is striking. Six kilometers of range between mesh points, enough to blanket a factory campus or a city district with a handful of devices. Up to ten megabits per second. Latency as low as one millisecond. And because there is no base station draining power, devices can run for years on a battery.
For the first time, there is a standard that does not force a trade-off between range, speed, power, and cost. But a standard is just a specification. It needs silicon to become real.
The chip that makes it real
Last Mile Semiconductor is building that silicon: the first chip to put an entire DECT NR+ radio, processor, AI engine, and security module onto a single piece of semiconductor. One chip, roughly the size of a fingernail, that gives any device everything it needs to join an NR+ network.
Putting a radio and a computer on the same chip sounds straightforward. It is not. The radio circuitry deals in continuous analog signals that are exquisitely sensitive to noise. The processor deals in digital logic that generates exactly the kind of noise the radio hates. Making them coexist on the same sliver of silicon, without one corrupting the other, is one of the hardest problems in chip design. It is also why this company exists in Dresden, one of the few cities in the world with a deep enough concentration of mixed-signal design expertise to even attempt it.
Other vendors have built NR+ solutions by wiring multiple chips together inside a single package. That works for prototyping, but it is too expensive, too power-hungry, and too large for the billions of devices that will eventually need this connectivity. Someone has to do the hard integration work of putting it all on one chip. That is what LMS is doing.
Why we believe this is the right time
When Texas Instruments paid $7.5 billion to acquire Silicon Labs earlier this year, it sent a clear signal: the company that owns the connectivity layer for IoT devices owns an enormously valuable position. The semiconductor industry is consolidating around that conviction.
LMS already has paying customers and a pipeline of large industrial and technology companies evaluating the chip for building automation, energy metering, industrial sensing, and professional audio. These are sectors where nothing on the market today does what NR+ does, and where the need is not theoretical.
Why Cloudberry VC
We invest in European semiconductor companies at the earliest stages, in the gap between the first working silicon and the first volume customer. This is where the risk is highest and where hands-on support matters most.
LMS fits our thesis precisely. A chip company in Dresden, sitting next to one of GlobalFoundries' most advanced manufacturing facilities, building on an engineering tradition that European universities have been cultivating for decades. We know this ecosystem because we invest across it.
Chip development is inherently iterative. You design, manufacture a test batch, measure what the simulation got wrong, redesign, and try again. LMS completed their first manufacturing run in August 2025 and the silicon is now in testing. More iterations will follow before volume production. We knew this going in.
What gives us conviction is that the deep, unrepeatable engineering work is done. The radio architecture, the signal processing design, the system model. Years of work that a new competitor cannot shortcut by hiring a bigger team. What remains is the execution of turning a working design into a production chip, and that is where our fund can help in ways most investors cannot. GlobalFoundries, the company manufacturing the LMS chip, is a partner in our fund. That relationship gives us direct visibility into the production schedule and a seat at the table when manufacturing decisions are made. In chip development, your relationship with your factory is as important as your own engineering. Having that connection is a strategic advantage we bring to every semiconductor company we back.
The founders
Christoph Gulich spent years building Deveritec, a semiconductor R&D services firm, before realising that the biggest opportunity was not in selling engineering hours but in owning the chip. He spun out Last Mile Semiconductor to do exactly that. Mario Orgis was part of the Intel XMM7160 modem team and supported the growth of another Dresden chip company from thirty to ninety engineers. He has taken mixed-signal designs from concept to production before, and that experience matters when you are building one of the most complex chips in the IoT space.
Together they have assembled a team of thirty engineers in Dresden. People who design radio circuits and analog chips for a living, in a discipline where the global talent pool could probably fit in a single conference room.