Briefing: AI Governance Is Splitting in Two — Only Half Is Secured

$9.75 billion. That's what the market committed to forward-deployed AI engineers in the past 12 months, per Tomasz Tunguz — spanning OpenAI's $4B DeployCo, Microsoft's $2.5B Frontier Company, and Anthropic's $1.5B Blackstone-backed Ode.

$30 billion. Apple and Broadcom's multiyear custom ASIC and AI server chip deal, securing the hardware layer of AI infrastructure.

Zero. The number of equivalent products, partnerships, or commitments addressing the governance of strategic AI decisions — what gets built, which tradeoffs to accept, and who owns the outcome.

The market just split AI governance into two layers. Your organization is likely only paying for one of them.

The Two-Layer Governance Split

On August 6, DXC and Primary launched an AI-native Zero Trust platform covering consulting, implementation, integration, governance, and operations for enterprise AI. It's a managed service — a packaged product — for infrastructure governance. You can now buy AI security, compliance, and data governance off the shelf.

That's Layer 1: infrastructure governance. It's becoming a product category. Providers are competing on it. Pricing is forming.

Layer 2: strategic decision governance — who decides priorities, how tradeoffs are evaluated, what outcomes define success, and who owns the answer — remains ungoverned. There is no product for it. There is no managed service. There is only the accountable human judgment layer.

The gap between these two layers is where AI spending produces infrastructure and stops short of strategic return.

What the Numbers Say

The adoption-to-impact gap is not a capability problem. It's a governance design problem. Organizations are spending on infrastructure governance (security, compliance, data access) while leaving strategic governance (tradeoffs, accountability, decision rights) to ad-hoc processes.

The Board-Ready Question

For every dollar your organization spends on AI infrastructure governance, how much is it spending on AI decision governance?

If the answer is "we haven't separated those," the governance architecture is incomplete. Infrastructure governance secures the pipes. Decision governance secures the strategy.

The market has already split. The providers have already chosen their layer. The question for the board is whether the organization has chosen — or is defaulting — into securing only half of its AI investment.