The Thesis
The cost of building fell faster than the cost of buying adapted, and this week's data shows enterprises are already acting on it.
The Signal
Three moves worth watching this week
Enterprises are replacing SaaS with custom software at scale.
Retool's 2026 Build vs. Buy report found that 35% of teams have already replaced at least one SaaS tool with a custom build, and 78% plan to build more internal tools this year.
Why it matters: AI-assisted development has compressed build timelines from months to days, and enterprise buyers are no longer captive to per-seat pricing inflation.
Second-order effect: Watch for SaaS vendors to accelerate platform lock-in through proprietary data formats and workflow dependencies as their moat erodes.AI interpretability graduated from research to funded product category.
Goodfire raised $150M at a $1.25B valuation for its model design environment, backed by B Capital, Salesforce Ventures, and Eric Schmidt. The company identified novel Alzheimer's biomarkers by reverse-engineering a foundation model with its interpretability tools.
Why it matters: As Colorado's AI Act takes effect in June and the EU high-risk AI rules approach, enterprises need auditability that goes beyond documentation.
Second-order effect: Watch for interpretability requirements to show up in enterprise procurement checklists and cyber insurance riders, creating a new compliance infrastructure layer.Capital concentration at frontier labs is approaching sovereign scale.
OpenAI is finalizing a $100B+ round at an $850B+ valuation, with Amazon, SoftBank, Nvidia, and Microsoft as lead backers. Anthropic closed a $30B round at $380B just days earlier, reporting $14B in annualized revenue.
Why it matters: The infrastructure cost to compete at the frontier now exceeds what all but a handful of sovereign-adjacent entities can finance, effectively capping the number of foundation model providers at three to five.
Second-order effect: Watch for the application layer to become the real competitive surface, as model access commoditizes and differentiation shifts to workflow integration and domain-specific data.
The Playbook
A Buy-vs-Build Rubric for AI Capabilities This Week
The Retool data and Constellation Research's 2026 enterprise trends analysis both point the same direction: AI-assisted coding has made "build" viable for capabilities that were buy-only twelve months ago. But viable is not the same as smart. MIT research cited in recent enterprise AI frameworks shows purchased AI solutions still succeed roughly 67% of the time vs. 22% for internal builds.
Use this five-point rubric before committing resources:
Differentiation test. Does this capability directly create competitive advantage that a vendor's other customers also get? If yes, build. If it's table-stakes functionality (auth, billing, basic analytics), buy.
Data gravity test. Does the capability require your proprietary data to work well, or does it work on generic data? High data gravity favors build. Low data gravity favors buy.
Maintenance budget test. Can you staff ongoing maintenance (API changes, security patches, model updates) for 24+ months? Retool's report found 60% of builders shipped software outside IT oversight last year. Shadow builds without maintenance budgets become technical debt.
Compliance surface test. Will this capability face regulatory scrutiny (hiring decisions, financial advice, health data)? If yes, the governance overhead of a custom build rises sharply. Consider buying the foundation and building only the differentiating layer, as Seekr's framework recommends.
Time-to-learning test. Is speed to market or speed to insight more valuable? If you need to learn what works before scaling, build a prototype. If you need production reliability in under 90 days, buy.
Score each dimension 1 (favors buy) to 3 (favors build). Total of 5–9: buy. 10–12: hybrid. 13–15: build.
The Verification Test
Claim: "Our AI agent handles 80% of [workflow] autonomously."
Test: Run 50 real edge cases from the last quarter. Measure completion rate, error rate, and fallback-to-human rate independently.
Pass criteria: Agent completes ≥40 of 50 cases end-to-end with <5% error rate and generates structured logs sufficient for a compliance audit.
Fail smell: Vendor demo uses curated examples only, declines to run on your data, or cannot produce an audit trail for any completed task.
HSI Note
Horizon Search Institute
Human Performance: Becker's survey of 103 health system leaders names human-AI integration as the top workforce challenge of 2026.
Responsible AI: Wilson Sonsini's 2026 preview notes SEC now recommends board-level AI governance disclosure tied to cybersecurity risk.
Planetary Futures: ISO released 14092:2026, a new global standard linking local climate adaptation planning to finance accountability.
Governance & Diplomacy: Trump's AI executive order proposes federal preemption of state AI laws, setting up a jurisdictional contest through 2026.
Links Worth Your Time
Retool: 2026 Build vs. Buy Report — Best primary data on how fast vibe-coding and shadow IT are reshaping enterprise software procurement.
Goodfire Series B announcement — The clearest articulation of why interpretability is an engineering discipline, not just a safety checkbox.
Constellation Research: 15 Enterprise Tech Trends for 2026 — Sharp analysis of why agentic licensing, build-over-buy, and physical AI are converging this year.
King & Spalding: State AI Laws vs. Federal Preemption — Essential legal briefing on the regulatory uncertainty facing every AI deployer in the U.S.
ISO 14092:2026 Climate Adaptation Standard — New global framework tying climate resilience planning to governance accountability and finance access.
The Builders & Doers Podcast
Alexander Harmsen breaks down how AI can improve financial outcomes without fully replacing humans, why personalization is where the real value is, and how he’s building in one of the hardest regulated categories.
Sources
Retool. "2026 Build vs. Buy Report." BusinessWire, Feb 17, 2026.
Goodfire. "AI Lab Goodfire Raises $150M at $1.25B Valuation." PRNewswire, Feb 5, 2026.
TechCrunch. "OpenAI reportedly finalizing $100B deal at more than $850B valuation." Feb 19, 2026.
CNBC. "Anthropic closes $30 billion funding round at $380 billion valuation." Feb 12, 2026.
Constellation Research. "Enterprise technology 2026: 15 AI, SaaS, data, business trends to watch." 2026.
Neontri. "Enterprise AI Roadmap 2026: Implementation Framework." Feb 2026.
Seekr. "Build vs. Buy AI: Why Enterprises Need a Production-Ready Foundation." Feb 10, 2026.
Wilson Sonsini. "2026 Year in Preview: AI Regulatory Developments." 2026.
ESG Today. "ISO Launches New Climate Adaptation Standard." Feb 2026.
Becker's Hospital Review. "10 healthcare workforce challenges defining 2026." Feb 2026.
Salesforce Ventures. "AI Needs Interpretability." Feb 5, 2026.
Bloomberg. "OpenAI Funding on Track to Top $100 Billion in Latest Round." Feb 19, 2026.
The Searchlight is a weekly field memo. It is not investment advice. Views are the editor's own.