stage 01 · build the third pillar

The Technology Model

Ask a room of executives how their technology makes money and you'll get a room's worth of different answers. That gap is the single most reliable place to find trapped value.

By · V-Squared AI

Ask a room of executives how their technology makes money, and you'll get a room's worth of different answers. Ask the engineers how the business monetizes what they build, and you'll often get a shrug. That gap, leaders who don't understand the technology model and technologists who don't understand the business model, is where value hides.

Three bottlenecks, one model

The technology model is the shared account of how technology works, why it works that way, and how it's monetized. It does three things at once. It removes the understand bottleneck, because everyone is finally looking at the same picture. It removes the work with bottleneck between the C-suite and technologists, because they share a vocabulary grounded in how value is actually created. And it sets up every align decision downstream, because you can't align a build to monetization that's unmapped.

Why the roadmap starts here

Strategy that isn't grounded in the technology model is guesswork that gets called planning, because it funds a picture of technology that doesn't exist. Build the shared technology model first, and opportunity discovery, maturity, and monetization all have something solid to stand on. This page is the linchpin between the Outcomes Economy's executive framing and the operational work below.

Strategy that isn't grounded in the technology model is guesswork that gets called planning.
the framework · outcomeseconomy.com

The executive framing sits atop the Outcomes Economy

This is the roadmap's first stage. The framework, and the three-bottleneck map above it, lives at the top of the pyramid.

Go to the Outcomes Economy
the evidence · earnings calls, july 2026

Microsoft wrote its technology model down and read it to investors.

This stage says you cannot align strategy to a model nobody can see. In July 2026 the largest enterprise software vendor made its own technology model explicit, in enough detail for a CFO to fund against, and the numbers came with it.

peer-reviewedsurvey researchcompany filingsmarket dataearnings call
Microsoftearnings callFY26 Q4 · Jul 29, 2026

How it gets built: keep the harness separate from the model

Nadella told enterprises to keep the harness separate from the model, so memory, context, and action space sit outside any model family and every model stays substitutable. Use frontier models where they earn it, cheap models where they do not, and train your own when you want neither, because you already hold the outputs, the traces, and the context. He said Microsoft intends to evangelize the design pattern.

Source: Microsoft FY26 Q4 earnings call, July 29, 2026.
what it confirms

This is the build half of a technology model, stated plainly enough to fund. The durable asset is the harness and the model is a substitutable input. An organization that cannot articulate that distinction keeps buying model access and wondering why nothing compounds.

Microsoftearnings callFY26 Q4 · Jul 29, 2026

What it costs: the routing ratio, published

A small cyber model outperforming a much larger frontier model at half the cost, with 90% of tasks handled by the small model and 10% escalated. An 89% GPU cost reduction in Dynamics 365 and up to 84% in PowerPoint. 10% lower median token usage on GitHub Copilot. Copilot workload throughput up 4 times since January. Maia 200 at 30% better performance per dollar.

Source: Microsoft FY26 Q4 earnings call, July 29, 2026.
what it confirms

Note what is being optimized. Not capability, but cost per delivered outcome. That is the number a technology model has to produce, and most enterprises cannot generate it because nobody has written down how the technology creates value in the first place.

Microsoftearnings callFY26 Q4 · Jul 29, 2026

How it makes money: token spend became a product

Amy Hood described the E7 SKU as selling both observability and manageability of token spend across business processes. Agent 365 registered close to 40 million agents across tens of thousands of companies in two months.

Source: Microsoft FY26 Q4 earnings call, July 29, 2026.
what it confirms

The monetization half of the model is no longer abstract. Consumption is the unit, governing consumption is itself a product, and an enterprise that cannot see its own token spend by business process has an unwritten technology model with an invoice attached.

Metaearnings callQ2 2026 · Jul 29, 2026

Same infrastructure, opposite conclusion

Zuckerberg said it would be foolish to simply sell all of the compute, and that he expects significantly higher margin from selling intelligence than from selling compute directly. He described running an efficient auction over compute the way Meta auctions ad inventory today.

Source: Meta Q2 2026 earnings call, July 29, 2026.
what it confirms

Two companies held comparable infrastructure and reached opposite monetization conclusions. The deciding input was an architectural read, not a finance one. That is exactly the decision an enterprise cannot make well when leaders and technologists hold different models of how the technology works.

Sources: Alphabet Q2 2026, Microsoft FY26 Q4, Meta Q2 2026, and Amazon Q2 2026 earnings calls and releases, July 2026. Figures are as stated by company executives. Amazon reported after market close on July 30, so Amazon figures come from the release and initial call remarks rather than a full transcript.

the wider evidence · the gap, measured

Money is being allocated from a picture of the technology that is wrong.

If the technology model were shared and accurate, funding would follow value. Three 2025 and 2026 datasets show it does not, and they locate the error at the level where the budget gets approved.

peer-reviewedsurvey researchcompany filingsmarket dataearnings call
MIT Project NANDAsurvey research2025 · 300+ deployments

The budget went where value was visible, not where it was largest

The GenAI Divide study reviewed more than 300 deployments with 52 interviews and 153 survey responses. Sales and marketing captured roughly 70% of AI budget allocation, while the study found the strongest returns in back-office automation across procurement, finance, and operations, largely through replacing outsourcing and agency spend. Around 95% of pilots produced no measurable P&L return, and the report names learning rather than infrastructure, regulation, or talent as the core barrier to scaling.

Source: MIT Project NANDA, The GenAI Divide: State of AI in Business 2025. The report's own exhibit puts sales and marketing near 70% of allocation; some secondary coverage cites 50%.
what it confirms

Money followed board visibility instead of value. That is what allocation looks like when there is no shared account of how the technology creates value, and it is why this stage comes before opportunity discovery rather than after it.

Boston Consulting Groupsurvey researchMay 2026 · 625 leaders

The CEO and the board are not funding from the same picture

BCG surveyed 351 CEOs and 274 board members at companies with at least $100 million in revenue. 61% of CEOs said their boards are rushing AI transformation. Around 75% of board members rated their own AI understanding at or above their peers, an assessment their CEOs did not share. CEOs estimated 35% of their performance evaluation depends on achieving AI return on investment, while boards put it at 27%.

Source: BCG, Split Decisions: The BCG CEOs and Boards Survey, May 2026.
what it confirms

Eight points of daylight on what the CEO is accountable for, and neither side is wrong, because there is no shared model to be wrong about. Every funding argument downstream inherits that ambiguity, which is why the roadmap cannot start at stage two.

Executive AI usagesurvey research6,000+ executives

Leaders are funding a technology they do not operate

A survey of more than 6,000 senior executives across four countries found nearly 70% of CEOs, CFOs, and senior executives use AI at work less than one hour per week, including 28% who never use it, while many of the same organizations set adoption mandates and track employee usage.

Source: survey of senior executives conducted with Stanford economist Nicholas Bloom, reported March 2026.
what it confirms

You cannot build an accurate model of how a technology creates value from a demo and a vendor deck. This is the understand bottleneck measured in hours per week, and it explains why funding decisions and engineering reality drift apart so reliably.

Boston Consulting Groupsurvey research10-20-70 allocation

The algorithm is the smallest part of the answer

BCG's allocation principle for deploying AI at scale puts 10% of the effort on algorithms, 20% on technology and data, and 70% on people and processes. The firm reports that organizations routinely invert the ratio, concentrating on the 10% while treating process redesign and adoption as a footnote.

Source: BCG, AI at Scale, 10-20-70 approach.
what it confirms

A technology model that only describes the algorithm describes 10% of the value. The other 90% is how it gets built into the business and how the business captures the result, which is precisely the account this stage exists to write down.

The NANDA study samples 300-plus deployments and the BCG survey covers companies above $100 million in revenue, so both describe the mid-market and enterprise segments rather than the whole market.