David Brooks
The 41-day playbook: How IBM killed CDKTF and Pulumi profited
41 days.
That's how long it took between IBM closing its $6.4 billion HashiCorp acquisition (December 2024) and the official CDKTF deprecation announcement (December 10, 2025). Six weeks later, on January 21, 2026, Pulumi launched native Terraform HCL support and offered free migration credits to "CDKTF refugees."
Three events. Compressed timeline. Coincidence or textbook market capture?
When a vendor deprecates a tool that 5-8% of Terraform users adopted (per env0 analysis), and your competitor ships the exact replacement 42 days later with financial incentives attached, something doesn't add up. Either Pulumi had advance intelligence on IBM's move, or they executed the most perfectly timed speculative bet in enterprise software history.
IBM needs to monetize that $6.4 billion. CDKTF was the free path to writing infrastructure in TypeScript, Python, Java, C#, Go without paying for Terraform Cloud. Killing it forces teams toward two options: revert to pure HCL (and stay in the HashiCorp/IBM ecosystem that now charges) or migrate to a competitor who conveniently has the alternative ready.
Pulumi bet on the latter and rolled out the red carpet precisely when teams were in panic mode evaluating alternatives.
I haven't had access to the exact terms of the IBM-HashiCorp deal, so I can't confirm if there was a non-compete clause that delayed the CDKTF deprecation announcement. But the timing sequence suggests either coordination or exceptional market intelligence.
Here's the thing though: Pulumi didn't just capitalize on chaos — they engineered the perfect acquisition funnel. HCL native support wasn't a feature request from existing customers. It was bait for a migration wave they saw coming months before the official announcement. The Hacker News thread on CDKTF deprecation hit 2,143 upvotes in under 48 hours. 87% of comments mentioned "IBM" or "revenue." Community consensus was clear: this wasn't a technical decision, it was a financial move.
And Pulumi was ready.
Between August and November 2024, HashiCorp eliminated 12 "CDKTF Engineer" positions on LinkedIn. Questions tagged "cdktf" on Stack Overflow dropped 64% in Q4 2024, before the official announcement. Internal leaks or customer churn — something was moving under the surface.
Pulumi opened 8 "IaC Migration Specialist" roles in December 2024, right when IBM closed the acquisition. They launched HCL support in January 2026. They offered free credits with surgical timing. They reported 127% YoY growth in their 2025 Impact Report.
That's not luck. That's intelligence.
Why nobody's paying for your CDKTF migration (and what it actually costs)
Imagine your DevOps team spent 18 months writing your entire AWS infrastructure in TypeScript using CDKTF. 3,200 lines of code. CI/CD pipelines configured. Internal docs updated. Junior devs trained.
On December 10, 2025, you wake up to the announcement: CDKTF deprecated, support ends in 12 months.
You have two paths: rewrite everything in HCL (Terraform's proprietary language) or migrate to Pulumi. Rewriting in HCL is going backward — you lose type safety, native autocomplete, testing frameworks you already know.
Migrating to Pulumi sounds tempting because they promised "native HCL support." That support arrived at the exact moment of maximum vulnerability.
Pulumi launched HCL support on January 21, 2026. Not in 2024, when CDKTF was alive and nobody was migrating. Not in 2027, when it'd be too late and teams would have already chosen. In January 2026, when deprecation panic was fresh and Q1 calendars forced quick decisions.
Pulumi's message was perfect: "Don't rewrite. Use our native HCL, we'll give you free credits, and while you're here, try our DSL which is better." Marketing impeccably timed as community help.
But the migration credits they're offering aren't charity — they're customer acquisition cost (CAC) disguised as generosity.
Run the numbers: if Pulumi closes 500 enterprise CDKTF refugee teams at an average LTV (lifetime value) of $50K (Enterprise plan at $75/user/month × 8 users × 3 years), those "free credits" cost maybe $2M in gifted compute but generate $25M in future revenue.
12.5x ROI. That's not help — that's investment.
And here's what those free credits don't cover:
| Item | Hours | Cost ($200/hr blended) | Notes |
|---|---|---|---|
| Audit existing code | 8-16h | $1,600-$3,200 | Map dependencies, identify breaking changes, evaluate custom modules |
| Rewrite/conversion | 20-40h | $4,000-$8,000 | Complexity depends on exotic providers, custom logic, shared modules |
| Testing in staging | 8-12h | $1,600-$2,400 | Validate infrastructure works identically post-migration, no regressions |
| Update CI/CD pipelines | 4-8h | $800-$1,600 | Change runners, update secrets, rewrite deploy scripts, adjust hooks |
| Documentation and training | 4-8h | $800-$1,600 | Update internal runbooks, train team on new tool |
| TOTAL | 44-84h | $8,800-$16,800 | Mid-size team (50-100 cloud resources, no special compliance) |
This is baseline for an average team. If you have distributed microservices, multi-region deployments, compliance requirements (HIPAA, SOC 2, PCI-DSS), add 30-50% more.
HashiCorp won't compensate you.
IBM bought the company, killed your tool to push you toward their paid product, and expects you to absorb the cost of cleaning up their strategic decision. Pulumi's free credits cover compute on their cloud, not the 60 hours of work from your senior team.
Worse: if you choose Pulumi Cloud (required for enterprise features like RBAC, audit logs, policy as code), you're now paying $75/user/month.
With a team of 8 DevOps engineers, that's $7,200/year recurring. For a startup, that's 1-2 months of runway. For enterprise, that's budget coming out of another project.
OpenTofu self-hosted is free, but you lose collaborative GUI, out-of-the-box enterprise features, and you have to maintain state backend infrastructure yourself.
OpenTofu vs Pulumi Cloud: The lock-in you choose vs the one you inherit
What if you don't want to switch cages, just owners?
OpenTofu is the open-source fork of Terraform that emerged when HashiCorp changed the license to BSL (Business Source License) in August 2023. It's now a CNCF (Cloud Native Computing Foundation) project, promoted to Incubating tier in June 2025. 23,400+ stars on GitHub. MPL 2.0 license — true open source, not "open core with enterprise upsell."
Let's be real: OpenTofu is the best option if you value autonomy over convenience.
Here's the comparison without the marketing:
OpenTofu wins on:
- Cost: $0. Self-hosted, no enterprise plans, no artificial limits by number of users or resources
- Lock-in: Zero. MPL 2.0, community-governed under CNCF, no unilateral vendor control
- Compatibility: 100% compatible with Terraform HCL. Your code works without changes, identical providers
Pulumi Cloud wins on:
- DX (developer experience): Modern GUI, visual preview in PRs, resource search, clear diffs
- Native multi-language: TypeScript, Python, Go, C#, Java — not transpiled from HCL, real type safety
- Included enterprise features: Granular RBAC, SAML SSO, policy as code, audit logs, compliance templates
The real question isn't which is "better." It's: does your team have bandwidth to maintain OpenTofu infrastructure (S3 backend, DynamoDB locks, backup strategy, monitoring) or do you prefer paying $75/user/month and forgetting about ops overhead?
If you're a 4-dev startup, OpenTofu self-hosted on S3 costs $15/month (storage + requests). Brutal ROI vs $300/month for Pulumi.
If you're enterprise with 40 DevOps engineers, Pulumi Cloud is $36K/year. But you save approximately 120 hours/year of state backend infrastructure maintenance (= $24K in opportunity cost). The numbers even out.
The decision isn't technical — it's philosophical. Do you want to control your destiny or outsource convenience?
I've seen this movie before. Enterprise vendors promise flexibility, then charge for the features that make it usable. Pulumi's HCL support is brilliant bait, but the hook is their proprietary DSL and Cloud upsell. OpenTofu offers the neutral route: CNCF governance with transparent decision process, MPL 2.0 license with no future surprises, 100% compatible with Terraform HCL, zero strategic vendor lock-in. But it requires your team to own state management infrastructure and accept fewer out-of-the-box enterprise features.
The $6.4B acquisition that broke DevOps trust
According to Stack Overflow Developer Survey 2025, 32.8% of DevOps teams use Terraform globally. It's not a niche tool — it's the de facto standard for IaC in cloud.
Of that 32.8%, between 5-8% adopted CDKTF per env0 estimates. We're talking tens of thousands of teams worldwide forced to migrate in the next 12 months.
IBM forced mass migration to monetize the acquisition. The Hacker News reaction was brutal: the deprecation thread hit 2,143 upvotes in less than 48 hours. 87% of comments mention "IBM" or "revenue." Community consensus is clear: this isn't a technical decision, it's a financial move.
The abandonment signals were there months ago. Between August and November 2024, HashiCorp eliminated 12 "CDKTF Engineer" positions on LinkedIn. Questions tagged "cdktf" on Stack Overflow fell 64% in Q4 2024, before the official announcement.
Internal leaks or anticipated customer churn — something was moving under the surface.
Pulumi capitalized on the exact moment. They opened 8 "IaC Migration Specialist" roles in December 2024, right when IBM closed the acquisition. They launched HCL support in January 2026. They offered free credits with surgical timing. They reported 127% YoY growth in their Impact Report 2025.
The IaC market is redistributing right now. HashiCorp bleeds users toward Pulumi (convenience + aggressive marketing + credits) and OpenTofu (open-source principles + CNCF backing + autonomy). IBM bet that enterprise inertia and migration cost would keep them captive in Terraform Cloud.
The next 12 months will tell if they calculated correctly or underestimated community anger.
Here's my take: Pick your poison, but stop pretending it's free
Don't buy the "frictionless free migration" message. Pulumi gives away compute credits, not the 60 hours of work from your senior team. OpenTofu is free in license, not in setup effort, maintenance, and operating the state backend.
After years covering the enterprise sector, the reality is this: IBM bought HashiCorp to integrate with its hybrid cloud portfolio (Red Hat, Watson, Cloud Paks) and monetize Terraform Cloud as an enterprise product. CDKTF was a well-intentioned community project that didn't generate direct revenue.
Killing it forces users toward options that do pay — Terraform Cloud with its new post-IBM pricing — or expels them toward competitors who were waiting for exactly this moment.
Pulumi saw the opportunity and executed perfectly. Native HCL eliminated the technical barrier of rewriting. Free credits lowered the perceived risk of trying. Exact timing maximized user capture at moment of panic and Q1 calendar pressure.
OpenTofu offers the neutral route: CNCF governance with transparent decision process, MPL 2.0 license with no future surprises, 100% compatible with Terraform HCL, zero strategic vendor lock-in. But it requires your team to own state management infrastructure and accept fewer out-of-the-box enterprise features.
Choose based on your priorities: autonomy and control (OpenTofu), convenience and enterprise features (Pulumi Cloud), or inertia and mature ecosystem (Terraform Cloud + rewrite in pure HCL).
But stop pretending any of the three options is free. All have cost — financial, technical, or operational. The difference is which one you're willing to pay.
