Lakshya

Security Engineering & Tooling · Chapter 5 of 8

The market — consolidation, pricing, and what it means for you

Two competitors, a 200x difference in outcome, and why every pricing model in this industry quietly shapes somebody's architecture.

5 min read1 diagramAll 8 chapters

You are not asked about market structure directly. You are asked why a tool was chosen, why a renewal went badly, or what you would do if a vendor is acquired — and the answers come from here.

What closed, and one correction

DealValueStatus
Google → Wiz$32BAnnounced March 2025, CLOSED 11 March 2026. Cleared unconditionally by the US, EU, Australia, Israel, Saudi Arabia, South Africa and Türkiye
Palo Alto → CyberArk~$25BAnnounced July 2025, CLOSED 11 February 2026
Cisco → Splunk$28BClosed 18 March 2024
Fortinet → Lacework$152.3MClosed 1 August 2024
Zscaler → Red Canary~$651MClosed August 2025
SentinelOne → Observo AI$225MSeptember 2025 — bought explicitly to decouple ingestion from the SIEM (the Security Information and Event Management platform that centralises logs)

The correction worth carrying: both Google/Wiz and Palo Alto/CyberArk are still routinely described as “pending” or “awaiting regulatory approval”. Both have closed. Google's public commitment is that Wiz products continue to work across all major clouds including its competitors' — whether that survives five years of Google Cloud sales incentives is the open question, and it is the one to raise if asked.

SAME CATEGORY. SAME YEARS. SAME PROBLEM STATEMENT. WIZ $32B Acquired by Google. Announced March 2025, CLOSED 11 March 2026 — cleared unconditionally in every jurisdiction. Google’s largest ever. LACEWORK $152M Raised $1.3B at an $8.3B valuation in Nov 2021 — the largest round in cybersecurity history at the time. Sold to Fortinet, closed Aug 2024. ~98% destroyed. THE DIFFERENCE WAS NOT TECHNOLOGY Lacework had strong ML-based anomaly detection. What it did not have was distribution, time-to-value, and the graph-based attack-path narrative that made a finding actionable. In security tooling, “detects more things” loses to “tells me which three things to fix”. This is the more useful of the two headlines, and almost nobody cites it.
Memorise this pair rather than the $32B headline. It answers “what makes a security product succeed” better than any feature comparison.

The private-equity layer, and what it predicts

One firm's cybersecurity portfolio alone spans Proofpoint, SailPoint, Darktrace, Ping Identity, Sophos, Imprivata and Exabeam — over $60B of enterprise value and roughly $8B of combined revenue. The playbook is roll-up, cost discipline, price increases at renewal, and eventual re-IPO or sale. That is not inherently bad — SailPoint was taken private and returned to public markets — but it predicts a specific pattern: slower innovation, harder renewals, and integration performed by acquisition rather than by engineering. If a tool you depend on is bought by PE, model a renewal price rise.

Does platformisation deliver?

Commercially, unambiguously yes. Next-generation security ARR growing 33% to $6.33B with record operating margins, while discounting aggressively, is the signature of successful bundling. The vendors are winning this argument.

Technically, partially. Real integration exists where products share a data plane. It is thin where products were merely co-sold. And the buyer's actual trade-off is rarely stated honestly by either side: platforms win on correlation, procurement overhead and total cost; best-of-breed wins on depth and replaceability. The under-priced risk of platformisation is concentration — one vendor relationship, one renewal negotiation with no leverage, one outage surface, and one integration to unpick if you leave.

Meanwhile the fragmentation platforms promise to solve is real and measured: an average of 10.9 consoles per security operations centre, with only 59% of tools pushing data to the SIEM. Note who commissioned that research, though — it was Microsoft, and its conclusion is “unify now”.

Pricing models, and the architecture each one distorts

CategoryModelWhat it punishes
SAST / SCAPer developer, committer or contributorLarge engineering orgs — and contractors and occasional committers, under a 90-day committer count
SecretsPer developer or per repoMakes monorepo-versus-polyrepo distort cost arbitrarily
CNAPPPer workload, resource or assetAutoscaling and ephemeral workloads — you pay for burst capacity. Container-dense and serverless estates hardest hit
EDRPer endpointServer fleets, VDI, ephemeral cloud instances
SIEMPer GB ingestedVerbose sources — VPC (Virtual Private Cloud) flow, DNS, CDN, EDR telemetry. It directly punishes the logs with the best detection value
DASTPer application or targetMicroservices — each service counts as a target

The observation that signals seniority

Every pricing model creates a perverse engineering incentive. Per-workload CNAPP pricing discourages decomposition. Per-GB SIEM pricing discourages logging. Per-developer AppSec pricing discourages giving developers access to security tools — the exact population you are trying to reach.

Notice when your architecture is being shaped by a pricing model rather than an engineering requirement. Most teams never name it out loud, and it is one of the cleanest ways to demonstrate that you have operated a programme rather than read about one.

Check yourself

Have Google/Wiz and Palo Alto/CyberArk closed?Yes — March 2026 and February 2026. Both are still widely described as pending.
What does the Wiz–Lacework contrast prove?Not technology. Lacework had strong ML detection. The difference was distribution, time-to-value, and the attack-path narrative. 'Detects more' loses to 'tells me which three to fix'.
What does PE ownership predict?Roll-up, cost discipline, price rises at renewal, and integration by acquisition rather than engineering. Model a renewal increase.
What is the under-priced risk of platformisation?Concentration — one relationship, one renewal with no leverage, one outage surface, one integration to unpick on exit.
Which pricing model punishes the best detection data?Per-GB SIEM pricing. The most verbose sources — VPC flow, DNS, CDN, EDR telemetry — are also the most valuable.
What is the senior observation about pricing?Every model creates a perverse engineering incentive. Notice when your architecture is being shaped by a price list rather than a requirement.
← Runtime, endpoints, and why SIEM economics are the architectureDoes any of it work? The evidence, honestly →