Naval just said the quiet part out loud, and the market already priced it in before most people noticed.

$2 trillion in software market cap evaporated between January 15 and February 14, 2026. Thirty days. Adobe, Microsoft, Salesforce, SAP, ServiceNow, and Oracle alone shed over $730 billion. The S&P 500 Software & Services Index is down 20% year to date while the broader market stayed relatively flat.

But Naval’s framing is too clean. The market isn’t saying “software is dead.” The market is saying “software that charges per seat is dead.”

Here’s why that distinction matters: If an AI agent does the work of three analysts, the customer doesn’t need three seats. Seat-based pricing was the foundational unit economics of SaaS for twenty years. Agents just broke the denominator. Revenue doesn’t decline because the product fails. Revenue declines because fewer humans touch the product. Atlassian is down 35%. Salesforce down 28%. Their core workflows, task tracking, data entry, customer logging, are exactly what agents automate first.

Meanwhile Palantir posted 70% revenue growth last quarter, guided for 61% in 2026, and trades at 44x forward sales. It grew revenue from 13% to 70% in ten consecutive quarters of acceleration. The reason Palantir survives the SaaSpocalypse is that it never sold seats. It sold decision infrastructure on top of messy, fragmented data that AI models can’t easily replicate.

The real split in the market: AI-native companies trade at a median 10x+ revenue multiple. Traditional SaaS sits below 5x. That gap was already wide six months ago. It doubled this year.

What Naval is picking up on is that the per-seat SaaS model was always renting access to a workflow. AI agents are replacing the workflow entirely. And once enterprises figure out that they can cut SaaS spend and headcount simultaneously, the compounding effect on license revenue gets ugly fast. Mizuho’s analyst put it bluntly: institutional buyers see no catalysts for a SaaS revaluation at any price.

The investable software companies in 2026 own one of three things: proprietary data that compounds, outcome-based pricing that scales with AI adoption, or infrastructure so deeply embedded that ripping it out costs more than keeping it. Everyone else is a melting ice cube charging monthly for the privilege.

Naftiko’s Take

How Naftiko Framework and Fabric Address Problem 4

The SaaSpocalypse article identifies three categories of software that survive the collapse of per-seat pricing. Naftiko addresses all three — and directly enables enterprises to navigate the SaaS market disruption from both sides.

Category 1: Proprietary Data That Compounds

The thesis: Palantir survives because it sells decision infrastructure on top of messy, fragmented data that AI models can’t easily replicate. Adobe, Salesforce, SAP are losing because their workflows — data entry, task tracking, customer logging — are exactly what agents automate first.

How Framework addresses it:

• Capabilities turn fragmented SaaS data into reusable, composable assets — instead of AI accessing Salesforce directly (and automating away the seat), a capability bundles the CRM fields + identity rules + PII policy into a governed unit. The enterprise’s combination of data sources, business rules, and access policies becomes the proprietary layer — something an AI model cannot replicate from the outside
• Format conversion (JSON, XML, Avro, Protobuf, CSV, YAML) — the messy, fragmented data problem Palantir solves with expensive professional services, Naftiko solves declaratively. Disparate SaaS data gets normalized into usable context without a re-platform
• Capability composition — aggregate capabilities roll up cross-system knowledge (CRM + ERP + identity + billing) into compound decision units. The value compounds with each new capability added to the fabric

How Fabric addresses it:

• consumesApis / providesApis relations in Backstage — the catalog makes the enterprise’s full integration graph visible and queryable. That map of what connects to what, with what policies, at what cost, is proprietary data that compounds over time. It can’t be replicated by a competitor or a new SaaS vendor
• Duplicate detection — surfaces overlapping capabilities as consolidation candidates. Over time the fabric becomes a curated, deduplicated knowledge base of how the enterprise’s systems actually relate — not how vendors claim they do
• Spec-as-source-of-truth — every label, catalog entity, and cost allocation entry is derived from the YAML spec. The spec library is the proprietary asset; it grows more valuable with every capability added

Category 2: Outcome-Based Pricing That Scales with AI Adoption

The thesis: Per-seat pricing breaks when an agent does the work of three analysts. Customers don’t need three seats. The denominator changed. Companies that sold outcomes rather than access are the ones growing.

How Framework addresses it:

• MCP and Agent Skills exposure — capabilities are the atomic unit of what agents invoke to produce outcomes. A capability isn’t a seat; it’s a function. Naftiko naturally aligns with outcome-based models because you measure what was done (operations invoked, data returned, orchestrations completed) rather than who logged in
• outputParameters with explicit contracts — the output of a capability is a defined, measurable artifact. Outcomes are observable and attributable by design, which is the prerequisite for outcome-based pricing models

How Fabric addresses it:

• Prometheus metrics per capability (request count, latency histograms, upstream error rates per namespace) — every capability invocation is measured. Usage data at the operation level is available without additional instrumentation. This is exactly the signal you need to price on outcomes rather than seats
• chargeable and metered tags on ConsumedHttpOperation — cost is tracked at the finest granularity. Enterprises can see the actual cost of each outcome delivered, enabling charge-back models by domain or business unit rather than headcount
• naftiko.io/cost-center label aggregation via Kubecost — outcome cost attribution by business domain is built into the deployment model. The infrastructure for outcome-based internal pricing exists from day one

Category 3: Infrastructure So Embedded That Ripping It Out Costs More Than Keeping It

The thesis: The third survivor category is infrastructure so deeply embedded that switching costs exceed value. Everyone else is “a melting ice cube charging monthly for the privilege.”

How Framework addresses it:

• YAML capability specs become organizational knowledge — the library of capability definitions encodes how the enterprise’s systems work together. That institutional knowledge is embedded in the spec files, not in a vendor’s proprietary format. It’s portable, version-controlled, and owned by the enterprise
• The spec is the source of truth, not the vendor — Naftiko’s design principle explicitly states this. The enterprise owns the integration layer; Naftiko is the runtime that executes it. Switching a consumed SaaS vendor means updating a baseUri and auth block in YAML — the capability contract stays intact
• NaftikoCapability CRD in Kubernetes — the capability is a Kubernetes-native resource, living in the platform GitOps repo. It’s as embedded as Kubernetes itself

How Fabric addresses it:

• Backstage catalog as institutional memory — every capability, its owners, its dependencies, its lifecycle stage, and its governance score lives in the catalog. That graph of organizational knowledge becomes harder to replace than any individual SaaS product
• lifecycle field on Exposes — the fabric tracks which capabilities are experimental, production, and deprecated. The enterprise has a living map of its integration maturity. No SaaS vendor provides this about their customers’ actual usage patterns
• GitOps-first deployment — capability specs live in version-controlled Git repos with full history. The embedded knowledge compounds over years of commits, just like Palantir’s data compounds over years of customer data ingestion

The Deeper Framing: Naftiko as the Enterprise Response to SaaSpocalypse

The article’s core observation is that AI agents are replacing SaaS workflows, not SaaS data. Seat-based pricing breaks when agents do the work. But the data, the integrations, the business rules — those still have to live somewhere.

Naftiko is what sits between the enterprise’s existing SaaS investments and the agents replacing the workflows that used to consume them:

Your data and APIs are not technical debt — they are your strategic inventory. The enterprises that survive the SaaSpocalypse are the ones that treat their existing integrations as the asset, not the SaaS vendors sitting on top of them. Naftiko is how you operationalize that position.