Dash0, an OpenTelemetry-native observability company, acquired Polar Signals, a Berlin startup that makes continuous profiling software. Polar Signals founder Frederic Branczyk announced the deal in a blog post on 17 August 2026.
The entire Polar Signals team is moving to Dash0, together with its profiler, its Great Lakes storage engine and the open-source Parca project. Neither company disclosed the price or other terms.
Polar Signals Cloud customers will see no change for now. Branczyk wrote that the product, data, profiler agent, user interface, MCP endpoint and support staff all stay the same, and that customers will be told directly and in advance if anything changes.
What Polar Signals built
Continuous profiling records, all the time and in production, where a program spends CPU and memory, down to the line of code. Polar Signals sold this as a hosted service, Polar Signals Cloud, and maintained Parca, an open-source profiler that uses eBPF to collect data with low overhead. The team has been merging parts of the Parca agent into the OpenTelemetry eBPF profiler and will keep maintaining Parca as open source.
The company also profiled GPUs. Branczyk said its work on continuous CUDA profiling in production, including program-counter sampling with stall-reason attribution, will continue at Dash0. In March 2024 Polar Signals said Vercel had chosen its cloud product for profiling Vercel's edge platform.
Before starting the company, Branczyk spent about ten years maintaining and working on Prometheus, and he worked in the CoreOS and Red Hat ecosystems, according to his post and to Lightspeed Venture Partners.
“We spent years building the best continuous profiler we could. Now we get to put it where it always belonged, next to everything else.”
A database built five times
Profiling data is hard to store because almost every stack-trace sample is unique, which produces unbounded cardinality. Branczyk wrote that nothing available off the shelf let every label become a queryable column, so the team built its own database. The first two attempts stayed prototypes, and the third and fourth, known as FrostDB, used columnar storage with Parquet and Apache Arrow but ran into design limits.
The fifth, Great Lakes, is the part Dash0 plans to use most widely. Branczyk said it will become the storage engine behind Dash0's SignalStore, replacing ClickHouse over time, so that metrics, logs, traces and profiles sit in one engine with one query language.
According to Branczyk, the deal began at a dinner in Berlin, where Dash0's Mirko raised the cost and performance limits of running ClickHouse at Dash0's scale and Branczyk mentioned the team's database work.
What Dash0 plans to do with it
Dash0 already combined metrics, logs and traces. Over the coming months, CPU, memory and GPU profiles will become a native part of that platform, so that a user can open a slow trace span and see what the processor was doing at the time.
Polar Signals listed two further features. Auto-Tune is a routine for Dash0's Agent0 AI agent that reads profiling data on a schedule and opens pull requests for optimisations, with the profiling evidence attached. SignalControl raises profiling detail automatically when something anomalous happens, so customers do not pay for that depth all the time. Dealroom reported that Dash0 framed the deal around software that is increasingly written and run by AI agents.
The founders and the funding
Polar Signals raised a seed round in 2021 co-led by Lightspeed and GV. In March 2024 it announced a further $6.8 million of seed money led by Spark Capital, with GV, Lightspeed, Haystack and Lorimer, which it said brought total funding to $10.8 million. Angel investors in that round included Guillermo Rauch, Julius Volz, Monica Sarbu, Jimmy Zelinskie, Nikhil Benesch and Erik Bernhardsson. Natalie Vais and Alex Polvi joined its board at that time.
Dash0 was founded in 2023, according to Dealroom, which reported that the company had recently raised $110 million in a Series B at a $1 billion valuation. Polar Signals did not disclose its headcount.
In his announcement, Branczyk thanked the investors who backed Polar Signals when it was little more than an idea, and the customers who ran an eBPF-based profiler in production at a time when, he wrote, using eBPF for anything other than networking still looked like an unusual choice.
