Pattern thesis: Every transformation era follows the same four-wave arc — infra enabler → tooling middleware → point solution explosion → control-layer consolidation. The biggest exits cluster in waves 1 and 4. Wave 3 sub-categories reveal the pattern most clearly: point solutions that own their data moat survive; those that sit on top of someone else's data get absorbed. We are in wave 3 of the AI era. Wave 4 is wide open.
Wave 1
Infra Enabler
Wave 2
Tooling & Middleware
Wave 3 · we are here
Point Solution Explosion
Wave 4 · open
Control-Layer Consolidation
Digital Transformation
~2000 – 2014
DevOps
~2010 – 2022
AI-Native
2022 – present
Wave 1 Infra enablers
AWS Heroku acq $212M Twilio IPO SendGrid acq $3B Cloudflare IPO
Cloud primitives and communication APIs. Companies could build without owning hardware. The land grab for developer trust begins.
↳ Payments infra
Stripe $50B val Braintree acq $800M Adyen IPO
The wave 1 sub-category that didn't get absorbed — Stripe built deep enough to become infrastructure itself. Lesson: own the data, not just the API.
→ Became standalone infra
↳ Identity & access
Okta IPO $25B peak Auth0 acq $6.5B OneLogin
Identity became mandatory compliance infrastructure. Okta won by being the neutral party — not owned by any cloud.
→ Became standalone infra
GitHub acq $7.5B Docker raised $340M HashiCorp acq $6.4B Puppet Chef
Source control, containers, config management. Dev and Ops still separate but the primitives for collapsing that wall are being laid.
↳ Secrets & config
Vault (HashiCorp) acq $6.4B 1Password $6.8B val Doppler
Secrets sprawled across .env files and CI configs. Vault became the standard. Doppler the dev-friendly layer on top.
Mistral $1B+ Cohere $445M xAI $50B val AI21 Labs Inflection acq Microsoft
Foundation model labs and inference providers. Raw capability layer. VC massively over-rotated here — most capital went to wave 1.
↳ Video & multimodal
Runway $1.5B val Stability AI Pika ElevenLabs $3.3B val
Foundation models for non-text modalities — video, image, audio, voice. Still early; no clear Stripe-equivalent has emerged yet.
→ Still fragmented
Wave 2 Tooling + middleware
MuleSoft acq $6.5B Segment acq $3.2B Apigee acq $625M Boomi acq $4B
Integration, API gateways, customer data pipelines. Legacy systems needed to talk to new SaaS. Boring infrastructure, enormous outcomes.
↳ CMS & content infra
Contentful $3B val Contentstack Sanity
Headless CMS decoupled content from presentation. Became foundational middleware for every marketing and ecommerce stack.
↳ eCommerce infra
Shopify IPO $200B peak BigCommerce IPO
Shopify won by becoming the operating system for commerce — not just a tool but a platform with its own ecosystem. The wave 2 company that ate wave 3.
→ Platform winner
CircleCI raised $315M Harness raised $425M JFrog IPO $4B Snyk $8.6B val
CI/CD, artifact management, security scanning. The pipeline became code. Deployments became reproducible. The build became observable.
↳ API management
Kong raised $200M Postman $5.6B val Stoplight Apigee → Google
APIs multiplied faster than teams could manage them. Gateway, testing, and documentation became separate tool categories — then started collapsing back together.
↳ Testing & quality
Cypress raised $65M BrowserStack $4B val Sauce Labs Mabl
Test automation lagged CI/CD adoption by years. Every team had a different framework. Consolidation is still incomplete.
LangChain raised $25M LlamaIndex Pinecone raised $138M Modal raised $67M Replicate
Orchestration, vector stores, inference infra. Developers needed ways to chain calls, store context, and run models at scale.
↳ GPU & compute infra
CoreWeave IPO $23B Lambda Labs raised $320M Vast.ai Crusoe raised $686M
GPU scarcity created a picks-and-shovels moment. CoreWeave's IPO is the wave 2 bellwether — same pattern as Rackspace in digital transformation.
→ Biggest wave 2 exit so far
↳ Evaluation & fine-tuning
Scale AI $13.8B val Braintrust raised $36M Patronus AI Promptfoo
Training data quality and model evaluation became the hidden bottleneck. Scale AI is the Segment of this era — owns the data pipeline before the model.
Wave 3 Point solutions
HubSpot IPO Zendesk IPO Dropbox IPO Veeva IPO $12B Slack acq $27B
Best-of-breed SaaS per function. Companies assembled 30–50 tool stacks. Integration debt compounded fast.
↳ Marketing automation
Marketo acq $4.75B Eloqua acq $871M Pardot acq $95M Klaviyo IPO
All three major players absorbed — Marketo into Adobe, Eloqua into Oracle, Pardot into Salesforce. Classic wave 3 fate: point solution with no data moat gets absorbed by the CRM that owns the customer record.
→ All absorbed by platform owners
↳ BI & analytics
Tableau acq $15.7B Looker acq $2.6B Qlik PE $3B Domo IPO
Winners on paper, absorbed in reality. Tableau → Salesforce, Looker → Google. The data layer got consolidated into cloud data platforms — Snowflake and Databricks were the real wave 4 winners.
→ Absorbed into data cloud platforms
↳ Customer success
Gainsight raised $220M Totango ChurnZero Mixpanel IPO
Churn became a board-level metric as SaaS companies matured. CS platforms created a new function — but still competing with CRM add-ons a decade later.
⚠ Still being squeezed by CRMs
↳ HR tech
Workday IPO $20B Greenhouse raised $255M Lever acq Lattice $3B val
Workday won by building the system of record. Recruiting tools stayed independent longer because hiring is a distinct workflow from core HR.
Datadog IPO $30B peak PagerDuty IPO Grafana $6B val Sentry raised $217M LaunchDarkly
Observability, alerting, feature flagging. Each team bought its own tool. Sprawl set in.
↳ Engineering intelligence
LinearB DevDynamics Jellyfish raised $71M GitPrime acq $170M Swarmia Allstacks
DORA metrics, cycle time, PR throughput — engineering visibility for VPs. Depended on data owned by GitHub, Jira, Linear. Those platforms built analytics natively and ate the category. Live proof point of the squeeze playing out.
⚠ Commoditised by platform owners
↳ Cloud cost management
CloudHealth acq VMware Apptio acq IBM $1.94B Spot.io acq $450M Infracost
Cloud bills exploded without visibility. All three major players acquired — none built a standalone moat. AWS/GCP/Azure cost explorer ate the basic tier.
→ Absorbed by cloud vendors and PE
↳ Incident management
OpsGenie acq Atlassian $295M Incident.io raised $62M FireHydrant Rootly
On-call and incident response became a discipline. OpsGenie absorbed by Atlassian; the newer generation building workflow and retrospective depth to avoid the same fate.
⚠ Mixed — some absorbed, some surviving
↳ DevSecOps
Snyk $8.6B val Aqua Security raised $265M Wiz $32B val Orca Security
Security bolted onto DevOps as an afterthought. Wiz won by being cloud-native from day one — the only sub-category to produce a wave 4-scale outcome from wave 3 timing. The exception that proves the moat rule.
→ Wiz the standout exception ($32B)
↑ We are roughly here
Cursor ~$9B val Harvey raised $300M Glean $4.6B val Writer raised $200M Sierra
AI copilots per workflow — coding, legal, sales, support, search. Fast adoption, minimal governance. Every team buying their own. The same wave 3 pattern, faster.
↳ Code AI
Cursor ~$9B val Windsurf acq Google $2.4B Tabnine Codeium GitHub Copilot
First and biggest wave 3 AI sub-category. Cursor won early mindshare. Windsurf's acquisition by Google at $2.4B is already the first live squeeze playing out — GitHub owns the IDE, the repo, and the developer relationship.
⚠ GitHub is the looming platform threat
↳ Vertical AI — Legal & finance
Harvey raised $300M EvenUp raised $135M Clio $3B val Rho Mosaic
Domain-specific AI with regulatory moats. Harvey's bet: legal workflows require domain depth that general models can't replicate cheaply. Strongest wave 3 moat if it holds — analogous to Veeva in the DT era.
→ Strongest moat thesis in wave 3
↳ GTM & sales AI
Clay $1.25B val 11x raised $24M Artisan Amplemarket Keyplay
Outbound and pipeline generation being rebuilt around AI agents. Clay built the data enrichment layer that everything else depends on — smarter positioning than pure workflow automation.
→ Fastest growing new category in 2025
↳ Customer support AI
Sierra raised $175M Intercom AI layer Aisera Forethought Decagon
First enterprise function to be fully re-platformed around AI agents. High churn risk for point solutions — Intercom, Zendesk, Salesforce Service Cloud all adding AI natively.
⚠ Incumbent platforms moving fast
↳ Healthcare AI
Abridge raised $150M Nabla raised $30M Amboss Suki AI
Clinical documentation and ambient AI for physicians. Regulatory barriers create moats. Abridge's UPMC partnership is the Veeva playbook — one major health system as a wedge into the rest.
→ Regulatory moat = durable wave 3
Wave 4 Control layer + consolidation
Snowflake IPO $120B peak Databricks $62B val ServiceNow IPO $100B+ Zapier $5B val
Data platforms and workflow orchestrators unified the chaos. The biggest exits of the era — and they came after the point solution sprawl, not before.
↳ Data governance
Alation raised $340M Collibra $5.5B val Atlan
Once data sprawled across 50 SaaS tools, governance became mandatory — compliance, lineage, access control. Late wave 4 but the pattern is clear: control always follows chaos.
Backstage Spotify OSS → CNCF Humanitec raised $30M Cortex Port Facets
Internal developer platforms abstracted K8s complexity. DevOps became platform engineering. The ops burden moved from individuals to a team with a product mandate.
↳ Cloud FinOps (mature)
Harness cloud cost module CAST AI Kubecost
Cloud cost management grew up into FinOps — a practice, not just a tool. Survived the wave 3 squeeze by embedding into the IDP layer.
? AI observability ? Agent orchestration ? Governance & audit ? Cost management
No clear winner. The control layer for AI agents at scale does not yet exist. This is the white space.
↳ LLM observability (early)
Arize AI raised $62M Helicone LangSmith Weights & Biases raised $250M Humanloop
Monitoring model calls, latency, cost, drift — the Datadog moment for AI. Still fragmented. No company has done to LLM ops what Datadog did to infra observability.
→ Earliest signal of wave 4
↳ AI governance & compliance
Credo AI Patronus AI Lakera raised $20M CalypsoAI Vanta expanding to AI
EU AI Act + enterprise risk mandates are forcing governance tooling. The Collibra of AI — boring, compliance-driven, and extremely durable once embedded. The most underinvested category right now.
→ Regulation is the forcing function
↳ Agent orchestration
Temporal $1.4B val E2B raised $13M Restack Superagent
As AI agents move from demos to production workflows, orchestration and state management become critical. Temporal has the head start — built for durable execution before AI agents were a category.
→ Pre-PMF, high conviction
Smart money signal
Sequoia, Bessemer bet on SaaS infra and integration middleware early. Biggest returns came from wave 4 — Snowflake, ServiceNow, Databricks. Wave 3 sub-categories that got absorbed (Marketo, Tableau) still produced strong outcomes — but for acquirers, not founders.
a16z, Sequoia backed GitHub, HashiCorp, Datadog in waves 1–2. Engineering intelligence attracted seed/Series A but no breakout — the category hit a data moat problem. Wiz ($32B) is the anomaly: security owned infrastructure data no one else could see.
a16z, Sequoia, Kleiner piled into foundation models and wave 2 tooling. Wave 3 point solutions getting funded reflexively. No one has clearly backed the wave 4 control layer at scale. The governance and observability category is 18–24 months from its Datadog moment.
The
Thesis
Wave 4 of the AI era is the largest unfunded surface area in enterprise software right now.
The sub-category graveyard in this table is the proof: marketing automation absorbed into CRM, BI absorbed into data clouds, engineering intelligence commoditised by GitHub — all wave 3 point solutions that lacked a data moat. The AI wave 3 is reproducing this pattern at speed across coding, GTM, legal, support, and healthcare. The companies that survive wave 3 either own regulated domain data (Harvey, Abridge) or become the control layer itself. Wave 4 — observability, governance, agent orchestration — does not yet have a clear funded winner. The company that builds the "Snowflake for AI workflows" or the "Backstage for agents" is not yet funded at scale. That is the bet.
Written by
Himanshu Saxena & Shubhanshu Singh
© Himanshu Saxena & Shubhanshu Singh. All rights reserved.