TOPSHOT - A robot using artificial intelligence is displayed at a stand during the International Telecommunication Union (ITU) AI for Good Global Summit in Geneva, on May 30, 2024. Humanity is in a race against time to harness the colossal emerging power of artificial intelligence for the good of all, while averting dire risks, a top UN official said. (Photo by Fabrice COFFRINI / AFP) (Photo by FABRICE COFFRINI/AFP via Getty Images)

AFP via Getty Images

Most enterprise AI strategies are still not fully developed. Companies hand their proprietary workflows, tribal knowledge, and competitive logic to closed-source model providers, mistaking token consumption for transformation. Chamath Palihapitiya verbalised it in a post this week: the real risk is that companies bleed their edge into a model while believing they are building one.

Palihapitiya founded 8090 in January 2024 with a stated goal of rebuilding enterprise software at 80% feature completeness for 90% less cost, combining AI tooling with offshore engineering. Software Factory, its flagship platform, is the operational answer to the knowledge-control problem he is describing. Usage, he noted in the same post, hit record highs in recent weeks.

The VC Angle: Governance as the New Moat

For investors, Palihapitiya's framing redraws the enterprise AI opportunity. The early thesis was productivity: AI compresses development cycles, reduces headcount per function, expands margins. That story has been hard to validate in the aggregate. U.S. labor productivity grew 1.2% in 2024 even as enterprise token consumption reportedly grew at triple-digit rates. The productivity premium from AI tools, at least in the short term, has not shown up at the macro level.

The second-order thesis, which 8090 is building toward, is control. Companies that deploy AI without governance frameworks are not gaining an edge; they are systematically externalizing one. OWASP's Top 10 for LLM Applications lists sensitive information disclosure as the number-two enterprise AI risk, directly behind prompt injection. Both are operational realities, not theoretical exposures. Samsung engineers pasting source code into ChatGPT and lawyers uploading confidential case documents are documented examples of employees treating closed-source models as trusted internal tools.

The investment implication is that governance infrastructure around AI deployment is becoming a distinct product category. EY's March 2026 partnership with 8090, deploying Software Factory across tens of thousands of consultants, is one signal. The deal is framed around legacy modernization and new product development, but the underlying value proposition is auditability: code that is documented, governed, and traceable.

The Knowledge Containment Problem

Palihapitiya's post was a direct response to a longer bear-case piece circulating under the label "The Big Rug," which argued that enterprise AI adoption is functioning as an involuntary training data pipeline for model providers. The core mechanism: employees use closed-source tools under mandate, their workflows and company-specific reasoning enter the model's context, and over time that tacit knowledge becomes embedded in inference behavior that the model provider owns.

Software Factory operates as a governed management layer: humans, agents, and AI collaborate inside a structure where code changes are documented, product and engineering plans stay synchronized, and institutional patterns are captured as repeatable "Assembly Lines" rather than dissipated into a third-party model's training corpus.

The distinction matters because the alternative to controlled deployment is not no AI. It is shadow AI. A 2026 Viking Cloud study cited by Vectra AI found that 97% of organizations reported GenAI security issues or breaches. FireTail's April 2026 analysis identified shadow AI, unsanctioned tools spreading department by department, as the most common entry point for data leakage. The gap between AI adoption and AI governance is not closing on its own.

What This Means for Founders and Investors

The market Palihapitiya is positioning 8090 to capture is the infrastructure layer that decides which companies keep their institutional knowledge proprietary as agentic AI matures. That is a large and largely unpriced opportunity.

For founders building in adjacent spaces, the architecture question is a big one. Platforms that route enterprise workflows through centrally owned model infrastructure face the same structural tension Palihapitiya describes. The products that survive the next phase of enterprise AI adoption will be the ones that let customers document and control their edge.

For investors, the framing suggests a rerating in category terms. The productivity tools story has not produced the macro numbers yet. The governance and control story, with real enterprise deployments, documented risk frameworks, and a major services partnership already in place, has a more legible path to durable margins. The question is whether the market has priced that distinction or is still conflating infrastructure spend with competitive advantage.