The end of the stack

Enterprise IT architecture exists to work around two constraints. Integration is expensive. Compute is centralised. The modern enterprise stack, from the data warehouse at the core to the integration middleware at the edge, is a 20-year response to those two facts. Both are dissolving. What replaces the stack is smaller, more federated, and more responsive than what it replaces.

The death of the data warehouse, the compression of middleware, the displacement of enterprise BI, the obsolescence of target-state architecture: each has been written about in isolation. They are the same event: they reinforce each other, and the compounding changes the timeline.

The central move

Intelligence has stopped being scarce. A model that reads a schema, writes a transformation, handles edge cases, and posts a result costs a few dollars per run and minutes to stand up. The economics of integration shift from capital project to commodity call.

This does not mean every enterprise transaction now flows through an LLM. It means the expensive thinking happens once, at design time, and the resulting deterministic pipeline runs cheap at volume.

The architecture splits into three layers.

• Design-time intelligence. The model reads the source system, understands the target, writes the mapper, identifies the exceptions, produces working code. One-off cost, bounded.
• Runtime determinism. The generated pipeline runs on ordinary infrastructure at ordinary cost. Same as any middleware today, without the consulting invoice that built it.
• Intelligence in the loop. Called only where it adds value: exception handling, interpretation of unstructured content, judgment calls, anomaly commentary. The model touches the 1% of events that need it, not the 99% that do not.

One caveat. The enterprise integration bill was never mostly the build. It was the run: connector currency as upstream APIs version and break, drift detection, the governed context an agent can be trusted to act on. Intelligence commoditised the build. It has not commoditised the trust. This is why the incumbents are embedding the model into the platform rather than losing the platform to it.

This is the shape. Everything else follows from it.

What I am not claiming

I am not claiming that existing enterprise systems disappear. SAP runs 77% of the Fortune 500. Oracle Financials, Workday, core banking platforms: these installed bases persist for a decade or more regardless of what happens at the architectural layer above them. The systems of record are not at risk. The integration, middleware, and analytical layers built on top of them are.

I am also not claiming this happens uniformly. The pattern is barbell. Experimental and edge workloads (management dashboards, internal tools, federated queries, operational reporting) move fast. Core transactional systems and regulatory pipelines move slowly, and for good reasons. Net new build tilts towards the new pattern; existing build persists.

The timeline question is therefore not "when does the old stack die" but "when does net new enterprise IT spend cross over from old pattern to new." My answer is 2027-2029 for most categories, later for regulated cores.

What breaks

Integration middleware. Gartner sized iPaaS at roughly $8.5bn in 2024, up 23%, and still forecasts $17bn by 2028 on a consensus growth path.¹ 2028 is closer to the peak than the base camp. But the compression does not arrive as a topline decline, and this is the correction the first eighteen months force. The forecast is holding because the incumbents are embedding the model into iPaaS (generated connectors, self-healing pipes) so the buyer does not build in-house. The displacement shows up instead as consolidation of the pure-plays. In November 2025 Salesforce closed its purchase of Informatica, one of the named top five, for $8bn, at a 34% discount to where the stock had traded, and framed it explicitly as the data foundation for agentic AI.² Absorbed, not killed. Expect the pure-play count to fall faster than the revenue line.

The former top five (Salesforce/MuleSoft, Oracle, Informatica, SAP, Boomi) captured about 58% of the market. With Informatica now inside Salesforce, that is four groups, and Salesforce's share steps up accordingly.

Data warehousing as the default. The Snowflake and Databricks thesis was: move all your data here, query it in one place. When queries can reach to the data in situ, through connector protocols and scoped access tokens, centralisation becomes optional. It remains correct for high-volume analytical workloads and model training. It stops being the right answer for operational reporting, management dashboards, and ad-hoc questions, which is most of what actually gets built. CIOs centralise the workloads that earn it and federate the rest.

Enterprise architecture as drawing exercise. Target-state diagrams, three-year roadmaps, canonical data models, architecture review boards. All artefacts of a world where rewiring was expensive and irreversible. When a new flow costs an afternoon, the correct posture is grow-as-needed, not plan-before-build. Architecture becomes editorial, not constructive: decide what is worth having, not what the topology should look like in 2029.

Management information software. Anaplan, Workday Adaptive, Oracle EPM, Tableau-plus-consultancy. These are thicker than dashboards: workflow, versioning, approvals, audit, scenario management. But they are increasingly workflow shells wrapped around a modelling capability that used to be scarce and no longer is. The shell survives. The six-figure premium attached to the scarce part compresses, as CFOs keep the governance and let their finance team maintain the models in-house.

The transformation project. Three-month consulting engagements existed because three months was the shortest credible delivery unit. When the work takes four days, the billing model breaks. The bill-by-hour advisory function survives only where the constraint is genuinely judgment, not delivery.

What survives and grows

Systems of record. More important, not less. If the query reaches to the source, the source must be correct, current, and well-governed. The incumbent ERP and CRM platforms find their moat deepening as federation becomes the access pattern.

Governance and guardrails. Security, audit, data residency, regulatory compliance, access control. These constraints do not dissolve. They matter more when the rate of new flows accelerates. The Chief Architect role converts to Head of Platform Engineering: someone who runs the guardrails and the observability, not someone who draws diagrams.

Judgment about what is worth building. The bottleneck moves from "can we build this" to "should we." Editorial capacity becomes the scarce resource. Strategy becomes a throughput problem.

Who profits

If the old stack compresses, somebody collects the newly available value. Three pools, and the ranking matters more than the list.

The trusted-context layer. This is the prize, and naming it correctly is the whole game. When intelligence is cheap and flows proliferate, the scarce complement is the correct, current, governed context an agent can be trusted to act against. Systems of record, data governance, observability, semantic and lineage layers: whoever owns that surface owns the rent. This is the recurring shape. As production is commoditised, scarcity migrates to whatever sits around production.

The systems of record. SAP, Oracle, Salesforce, Workday. Their data becomes the queryable surface the federated architecture runs against. Clean APIs and well-documented schemas become a competitive moat they already partly have.

The model providers. Anthropic, OpenAI, and whoever else clears the frontier bar. They capture the design-time inference, which is real but a deflating pool: the intelligence is cheap precisely because it is no longer scarce, and inference prices fall as fast as capability rises. Commoditised layers do not collect the rent. This is the pool most likely to be over-estimated, not under.

The surprise of the first eighteen months is not that the pure-plays died. It is that the incumbents bought the trusted-context layer before the market repriced it. Salesforce took out Informatica at a discount and called it the data foundation for agentic AI. Fivetran and dbt merged into the same layer for agents.³ The value migrated to the trusted-context layer, captured by consolidation, not by displacement, and not by pure-play governance vendors.

Federation between companies

Inside a company, data sharing is a political problem dressed as a technical one. Between companies, it has been genuinely technical. No trust model, no standard for ephemeral access, no way to expose a narrow query surface without a bilateral integration project.

The pieces are now arriving on schedule. MCP-class protocols provide the wire format. OAuth-scoped tokens and verifiable credentials provide the trust model. Just-in-time connectors mean the pipe is stood up for the decision, used, and torn down. An open standard for exactly this, an agent-facing shared context schema, shipped in 2026.⁴ A supplier's inventory planning agent queries your production forecast at the moment it matters, scoped to the SKUs it buys, with a 24-hour access window. No shared database. No master agreement beyond a standing trust relationship.

The format was never the hard part between companies. Willingness is. A supplier exposing a query surface onto competitively sensitive inventory faces the same political and liability problem that stalls sharing inside the firm. The pipe being cheap does not make the counterparty want to open it. So the technical unlock lands first where incentives already align (supply chain partners with shared upside, correspondent networks, payer-provider exchange) and stalls where they do not. Supply chain visibility, ecosystem-level intelligence, federated benchmarking: buildable for the first time, because the transaction cost of narrow, time-bound, policy-bound data sharing drops to near zero. Built, only where somebody wants the sharing to happen.

The clock changes too

The quarterly beat of enterprise IT existed because that was the shortest unit in which anything could happen. Scope, build, test, deploy. A year to know if an idea was worth having.

The new clock is days. Hours at the edge. Teams that internalise this run fifty learning cycles for every one their slower competitors complete. The gap between a fast organisation and a slow organisation stops being 2x and becomes 20x, because compounding. The architecture diagram matters less than the metabolic rate of the organisation using it. The correct response for an operator is not to wait for the architecture to settle. The architecture will not settle for a decade. Run personal experiments at the new clock speed, immediately, against real problems. The scarcest knowledge in any organisation is direct, recent, first-hand experience of what this stack can actually do on a Tuesday afternoon.

The case against

Three serious objections.

Enterprise inertia wins for longer than bulls think. Banks still run COBOL. Large enterprises maintain technology a generation behind frontier for reasons that include risk management, audit requirements, vendor relationships, and institutional capacity for change. The new stack may be strictly better and still fail to displace the old one inside the decade. Counter: inertia moves the dates, not the direction of travel. The barbell holds regardless.

The reliability gap is real and under-acknowledged. LLM-generated integrations have a known failure mode: silent drift. The source schema changes, the generated mapper keeps running, the output is subtly wrong for weeks before anyone notices. This is not a theoretical risk. It is the argument against using this pattern for anything that touches money or regulation. Counter: wrap the generated pipeline in deterministic test suites and schema contracts; the pattern itself is not the problem. But it does mean the technology stays on the edges longer than the architecture optimists claim, and it is the reason the market is paying up for the trusted-context layer rather than the build.

Regulated industries have veto power. GDPR, HIPAA, banking secrecy laws, export controls do not care whether the data moved or was queried in place. A federated query returning personal data is still a data transfer. The legal scaffolding required to make cross-company federation routine is a 5-10 year project at regulatory pace. Counter: accepted. Regulated cross-company federation lands in year 5-10 and only in jurisdictions that explicitly bless the model. Internal federation and non-regulated cross-company federation land much sooner.

Falsifiable predictions

1. Revised (Jul 2026). By end-2027, at least two of the 2023 top-five iPaaS pure-plays cease to exist as independent public companies. The original prediction, a 20%+ downgrade to the $17bn 2028 forecast, is tracking against me: Gartner reaffirmed the forecast in 2025. The mechanism was wrong. Compression is arriving as consolidation, not as a cut topline. Informatica is already gone.
2. By end-2028, new-build corporate data warehousing projects decline 30% from 2024 peak as federated-query patterns displace them for operational and management-reporting workloads. Snowflake and Databricks pivot successfully into compute-for-inference. Fivetran and dbt do not. Status (Jul 2026): live and looking weaker. Fivetran and dbt merged in 2025 and are attempting exactly the agent-data-foundation pivot this predicts they will fail. Too early to score, but the sub-claim is now the exposed one.
3. By end-2029, the Enterprise Architect role as currently defined shows 40% or more decline in job postings from 2024 peak, replaced by smaller, sharper platform-engineering roles.
4. By end-2027, at least one top-five global systems integrator (Accenture, Capgemini, Infosys, TCS, Wipro) publishes a material restructuring of its integration practice, framed as "AI-accelerated," read as managed decline. Status (Jul 2026): hit, roughly two years early. Accenture folded strategy, consulting, technology and operations into a single "Reinvention Services" unit in mid-2025, on an $865m programme, with about 22,000 roles exited on a compressed timeline.⁵
5. By end-2030, federated queries between non-competing companies are a standard pattern in at least three industries: supply chain, financial services correspondent networks, healthcare payer-provider data exchange.

Any one of these failing to land inside a year of the stated window weakens the thesis. Three or more failing invalidates it.

Close

The architecture of the future is being built by the people who stop planning it and start shipping it. The firms that win the decade will not be the ones with the best architecture. They will be the ones that learn at the fastest clock speed. Everything else is legacy defending ground.

Notes

1. iPaaS grew 23.4% to $8.5bn in 2024 per Gartner's Market Share Analysis: Integration Platform as a Service, Worldwide, 2024; the $17bn-by-2028 forecast was restated in the Gartner Magic Quadrant for iPaaS, 19 May 2025. Gartner attributes continued growth partly to generative AI being embedded into iPaaS platforms. ↩
2. Salesforce completed its acquisition of Informatica on 18 November 2025 at $25 per share, an $8bn equity value, positioned as "the data foundation for agentic AI" (Salesforce press release, 18 Nov 2025). The $25 price was roughly a 34% discount to Informatica's trading level when acquisition talks were first reported (Salesforce Ben, 19 Nov 2025). ↩
3. Fivetran and dbt Labs announced an all-stock merger on 13 October 2025, combined ARR approaching $600m, completing in 2026, positioned as "the data infrastructure for trusted AI agents" (Fivetran press releases, Oct 2025 and 2026; TechTarget, 2026). Commentators read the deal as a survival and IPO-readiness move as much as a product strategy. ↩
4. "Agents Schema," an open standard designating a shared context schema in a warehouse or lake for agentic AI, was released alongside the Fivetran + dbt merger completion in 2026 (Fivetran / dbt Labs, 2026). ↩
5. Accenture announced its "Reinvention Services" reorganisation on 20 June 2025 (Accenture Form 8-K), consolidating Strategy, Consulting, Song, Technology and Operations into one unit, and detailed an $865m, six-month optimisation programme with roughly 22,000 roles exited across FY2025 (Reuters / CNBC, Sep 2025). ↩