Dash0 Acquires Berlin Software Startup to Build Systems That Heal Themselves

New York-based Dash0, fresh off a $110 million Series B that vaulted it to unicorn status, has acquired Berlin-based Polar Signals in a move that repositions the startup from a monitoring tool into a platform where AI agents detect, diagnose, and fix code failures without human intervention.

The deal, announced August 17, 2026, brings together Dash0's existing observability stack with Polar Signals' continuous profiling technology and a proprietary storage engine called "Great Lakes." Financial terms were not disclosed. But the strategic logic is clear: Dash0's 750-plus enterprise customers, including Taco Bell and The Telegraph, need infrastructure that does more than flag problems. They need software that resolves them. Founder and CEO Mirko Novakovic put it plainly. "Software is increasingly being written and operated by agents," he said, and those agents require precise information about "what code actually does, down to the individual line of code, the kernel, and the costs it generates."

The acquisition arrives as developer infrastructure enters a consolidation cycle shaped by AI agent deployment. The same week Dash0 announced the Polar Signals deal, Temporal Technologies entered talks to raise approximately $500 million at a valuation of at least $12 billion, more than double the $5 billion it reached in February 2026 in a $300 million round led by Andreessen Horowitz. Across the stack, from workflow orchestration to observability to silicon, capital is chasing the same thesis: AI agents fail in complex, silent ways, and the companies that prevent or repair those failures stand to capture significant enterprise spend.

What could happen next follows directly from this capital concentration. If Dash0 can deliver on autonomous code repair at scale, the company moves from competing with Datadog and New Relic on dashboards to displacing them on value. That repositioning, if successful, would support a substantially higher exit multiple than traditional observability software commands today.

Polar Signals Is Not a Feature Addition. It Is an Infrastructure Rebuild.

Continuous profiling gives developers visibility into which functions and lines of code consume CPU and memory. Polar Signals built that capability on top of a custom-designed storage engine called "Great Lakes," engineered to handle high-volume profiling data at scale. Dash0 is not simply adding a new screen to its platform. It is replacing ClickHouse, its existing database layer, with Great Lakes as the foundation for a rebuilt component called SignalStore.

This distinction matters for investors. A feature acquisition typically justifies a small premium over team cost. An infrastructure acquisition, one that rewrites the data layer underpinning a platform serving 750 enterprise customers, signals the company is building toward a significantly different architecture than the one it shipped at Series B. The engineering team from Polar Signals transfers with the deal, which means Dash0 acquires both the intellectual property and the people who built and understand it.

The scale of the infrastructure bet also explains why Dash0 moved quickly after its Series B rather than waiting. The $110 million raise gave it the balance sheet to absorb an acquisition without diluting the engineering roadmap. Moving the storage layer now, before customer volume grows further, is cheaper than migrating later.

Agent0 and AutoTune: The Revenue Thesis Behind the Technology

Dash0's planned agentic capabilities, operating under the name Agent0, represent the commercial logic of the acquisition. The company is developing a feature called AutoTune, which uses profiling data to identify inefficient code, generate improvement suggestions, and prepare pull requests automatically.

The workflow is specific: AutoTune detects CPU or memory inefficiencies, proposes fixes, and stages them for developer review. It does not deploy autonomously, at least not in the disclosed version. But the direction is unambiguous. Dash0 is building toward a system where the observability platform becomes an active participant in the software development cycle, not a passive reporter.

This mirrors the trajectory Temporal is following from a different angle. Temporal's orchestration platform keeps multi-step AI agent workflows from failing midway through execution. Its biggest customer is OpenAI, and the company also works with Replit and Abridge. Temporal keeps agents running. Dash0, with Agent0, intends to make agents smarter about the code they run. These are adjacent, complementary problems. Neither company competes with the other today.

Novakovic's Playbook: Instana, IBM, and the Serial Founder Discount

Mirko Novakovic is not building his first observability company. Before Dash0, he founded Instana, an application performance monitoring startup that IBM acquired in 2020 as part of IBM's push into cloud and application management. Terms of that acquisition were not publicly disclosed, but the deal established Novakovic's credentials in a market that includes Datadog, which trades publicly, and New Relic.

Novakovic founded Dash0 in 2023, three years after the IBM-Instana deal closed, and brought several former Instana engineers with him. The investor syndicate reflects that credibility. Balderton Capital, Accel, Cherry Ventures, DTCP, DIG Ventures, July Fund, and T.Capital participated in the $110 million Series B that established the $1 billion valuation.

The serial founder dynamic compresses the validation cycle. Novakovic does not need to prove the market exists. He built and sold a company in it. What Dash0 must prove is that the AI agent layer creates a new ceiling for what observability software is worth, and that Dash0 can reach that ceiling before Datadog builds the same capability organically or acquires a competitor.

The Power Layer Closes the Loop

The infrastructure thesis extends below the software stack. Velaura AI, a Santa Clara-based chip designer, raised $110 million at a valuation above $1 billion in a Series A led by Seligman Ventures, announced August 19, 2026. The company claims its Titan Core platform delivers two to four times the performance per watt for AI workloads, without reducing throughput. Velaura licenses its technology under an Arm-style model: an upfront fee plus a royalty tied to actual power savings a customer achieves. CEO Rajiv Khemani told Reuters the company is in talks with three of the four largest cloud providers, though he declined to name them.

Velaura's designs are stated to run in more than 30 million chips today. The company has backing from Samsung Catalyst Fund, Mayfield, Capricorn Investment Group, Prosperity7 Ventures, Premji Invest, StepStone Group, Maverick Silicon, and MARA, the bitcoin miner that has moved into AI computing.

The connection to Dash0 is indirect but structural. AutoTune's stated goal is to reduce CPU and memory usage. Velaura's stated goal is to reduce the power those CPU cycles consume. Both products target the same economic pressure: AI agent workloads are expensive to run at scale, and the enterprises deploying them are looking for every efficiency lever available.

The Plocamium View

The market is framing Dash0's Polar Signals acquisition as an observability product expansion. That framing undervalues what is actually happening.

Novakovic is not building a better monitoring dashboard. He is building the connective tissue between AI-generated code and AI-operated infrastructure. Agent0 and AutoTune are early-stage, but they define a category: agentic DevOps, where the software development cycle runs with minimal human intervention across write, deploy, monitor, diagnose, and repair.

The second-order implication is a compression of the observability market as currently defined. If Dash0's platform eventually closes the loop from detection to remediation, the value of a tool that only detects shrinks. Datadog, currently the dominant publicly traded player in observability, has not disclosed a comparable agentic repair capability. That window will not stay open indefinitely.

The Temporal data point is the corroboration. A valuation that doubled twice in 2026, reaching $12 billion in talks, reflects what institutional capital assigns to infrastructure that keeps AI agents running reliably. Dash0 at $1 billion is either dramatically undervalued relative to that benchmark, or it must prove the AutoTune roadmap ships at enterprise scale before the next funding round prices the gap.

Our thesis: the acquisition of Polar Signals, specifically the Great Lakes storage engine and the profiling team, is the technical prerequisite for Dash0 to compete in the same valuation tier as Temporal within 18 to 24 months. The storage layer is not a side bet. It is the foundation. SignalStore, built on Great Lakes, has to handle the data volume that Agent0 will generate at scale across 750-plus enterprise customers. Without it, AutoTune cannot function. With it, Dash0 has a defensible technical moat that pure-monitoring competitors cannot replicate quickly.

The PE entry point is pre-Series C, if one is forthcoming. The $110 million Series B at $1 billion implies a post-money valuation that looks conservative against Temporal's trajectory. The risk is execution: storage migrations are difficult, agentic capabilities are unproven at enterprise scale, and Datadog has both the capital and the customer base to compete aggressively.

The Bottom Line

Dash0's acquisition of Polar Signals is the first concrete move in a longer campaign to make autonomous code repair commercially viable at enterprise scale. The $110 million Series B gives Novakovic the runway to ship it. The 750-customer base, anchored by names like Taco Bell and The Telegraph, gives him the deployment surface to prove it. The Instana-to-IBM precedent gives investors confidence he knows how to build, scale, and exit in this market.

The structural question for institutional capital is not whether AI agents need better observability. They do, and the Temporal fundraise at $12 billion confirms the market knows it. The question is whether Dash0 can move from monitoring the problem to solving it automatically, before the incumbents build or buy the same capability. Novakovic has done this once. The evidence suggests he is building the technical prerequisites to do it again at a higher multiple.