8/647 → 29/1165; +1.25 percentage points of formation share.
The same 2020 cohort that built pipes to move data around has quietly split into two different businesses: companies like Polytomic, Jitsu, Mozart Data, and Prequel are still solving the mechanical problem of getting data from one warehouse or SaaS tool to another, while Metaplane, Chaos Genius, Monosi, Streamdal, and Datafold have moved on to the harder question of whether that data can be trusted once it arrives, watching for the freshness failures and schema drift that used to surface only when someone in finance noticed a broken number. That shift toward observability and lineage tracking lines up closely with the live concern about dependency graphs where application behavior now hinges on data, labels, prompts, and models whose changes go untracked, and with the related worry about giving AI agents durable provenance from evidence to conclusion. Airbyte and Operator.io mark a further turn, repositioning existing sync infrastructure as the context layer that AI agents query before acting, which only became a plausible product once MCP-style tool use and agent frameworks matured in 2024. This formation's own 2.01x growth looks almost modest next to Databases at 4.27x and Big Data at 3.60x, suggesting the real acceleration is happening one layer down, in the storage and processing substrate, while data engineering tooling is racing to keep pipelines and trust checks from becoming the bottleneck underneath it.
Agents that propose experiments or scientific conclusions need durable lineage from source evidence through code, parameters, instruments, intermediate results, and revisions.
Compare scientific-agent workflows with current ELN, data-lineage, model-tracking, and reproducibility systems to locate evidence breaks and manual handoffs.
Application behavior increasingly depends on data, APIs, labels, prompts, and models whose changes are not represented in a single operational dependency graph.
Map production incidents to upstream data, API, labeling, and model changes, then test whether teams can predict affected behavior before release.
With Metaplane, Datafold, Monosi, Streamdal, and Chaos Genius all independently building data quality/monitoring tools, teams adopting multiple pipeline tools (Airbyte, Polytomic, Supabase) will need a unified layer to reconcile alerts and lineage across these overlapping observability point-solutions.
Check back around August 2027 for new YC batches, to see whether independent evidence for "Data observability aggregation/orchestration layer" has appeared.
The proliferation of ELT and monitoring tools (Airbyte, Polytomic, Metaplane, Datafold) run in parallel across a company's stack will drive unpredictable compute/storage costs, creating an opening for tools that track and optimize spend across these fragmented data pipelines.
Check back around November 2027 for new YC batches, to see whether independent evidence for "Data pipeline cost/usage governance tools" has appeared.
Dataland's focus on data visualization plus multiple ELT tools (Polytomic, Airbyte) feeding disparate warehouses implies a fragmentation problem where metrics get redefined per tool; a shared semantic layer becomes necessary to keep dashboards and quality checks consistent.
Check back around November 2027 for new YC batches, to see whether independent evidence for "Semantic/metrics layer standardization" has appeared.
Included in YC's Data Engineering membership during the measured recent window.
Included in YC's Data Engineering membership during the measured recent window.
Included in YC's Data Engineering membership during the measured recent window.
Included in YC's Data Engineering membership during the measured recent window.
Included in YC's Data Engineering membership during the measured recent window.
Included in YC's Data Engineering membership during the measured recent window.
Included in YC's Data Engineering membership during the measured recent window.
Included in YC's Data Engineering membership during the measured recent window.
Included in YC's Data Engineering membership during the measured recent window.
Included in YC's Data Engineering membership during the measured recent window.
A primary internal company segment within Data Engineering, derived deterministically from the shared yc tag “Developer Tools”. This is a classification of current window members, not a causal layer or a separate time trend.
A primary internal company segment within Data Engineering, derived deterministically from the shared yc industry “Analytics”. This is a classification of current window members, not a causal layer or a separate time trend.
Measured-window members of Data Engineering that could not be assigned to a narrower shared YC tag or industry without forcing a classification.
A primary internal company segment within Data Engineering, derived deterministically from the shared yc industry “Engineering, Product and Design”. This is a classification of current window members, not a causal layer or a separate time trend.
No comparable outside activity is connected yet.
These share a word with “Data Engineering” but are tracked as separate categories. The real overlap below is the actual share of Data Engineering’s companies also found in that category — not a guess.
Cambrian Signal, “Data Engineering,” publication yc-2026-10-02-semantic-v8, snapshot October 2, 2026.
cambrian.jafar.com/formation/data-engineering