Strategy

AI Makes Prototypes Cheap. Three Years of Ownership Still Costs More

AI makes prototypes dramatically cheaper, but operating a capability for years still demands money and people. A three-year test and four criteria sharpen build, buy and partner decisions.

By Olivia Hart

5 min read

Your company needs a new capability. Should you build it, buy it, or partner for it?
Your company needs a new capability. Should you build it, buy it, or partner for it?Pexels / pixabay

What's News

  • AI makes prototypes dramatically cheaper but reduces the cost of operating a capability far less.
  • The recommended planning horizon is three years, covering building, maintaining and improving a capability.
  • Executives should compare their last three internal builds with original time, cost and staffing estimates.
  • Four tests shape the choice: defensibility, time to value, integration risk and cultural fit.
  • Build, buy and partner routes can be combined or changed as the company learns.

By Olivia Hart

AI has made prototypes dramatically cheaper, but it cuts the three-year cost of owning and operating a capability far less. The strategic mistake is to confuse a cheaper launch with an affordable commitment that can last three years.

Years ago, I urged the leadership of a global consumer electronics company to acquire a relatively young streaming-device company. The TV screen was becoming a dumb pipe, while value moved to the software that decided what viewers watched. Owning the startup would let the company keep making world-class TVs and extend its reach onto screens it did not make.

Internal teams disagreed. They had deep hardware expertise, strong engineers and their own smart TV software. Their logic was that buying was unnecessary because the company could build the product.

A few years later, the company shipped its own streaming device. It was not as good as the startup’s device, never gained traction, and the project was shut down. The startup went public and became a leader in streaming.

The missing asset was not a device the company could assemble. It was the market position the startup already had.

Why do leaders keep defaulting to build?

AI has accelerated the pace at which businesses can develop products, enter markets and reach customers. It also exposes gaps in corporate capabilities, forcing leaders to revisit the build, buy or partner decision more often. Many are defaulting to build.

Three biases distort that choice:

  • Overconfidence: One version assumes four people and six months will be enough.
  • Not-invented-here syndrome: The company treats outside capabilities as inferior to its own engineering.
  • Launch bias: Management estimates what it takes to launch a product, not what it takes to run the capability for years.

AI overconfidence adds another distortion. Leaders assume AI will sharply reduce both the cost and effort of building in-house. The savings are real, but they are front-loaded. AI makes prototypes dramatically cheaper, yet it shrinks the cost of maintaining, securing and improving a capability far less.

Internal teams also lose the people they expected to have. Employees get reassigned as other priorities take hold, and leaders confuse a faster launch with effective ownership. By then, a competitor may have reached the market or the project may have lost internal support.

Cheap prototypes do not erase the cost of running a business.

What should leaders check before spending?

Ryland Grace, Ryan Gosling’s character in Project Hail Mary, offers a useful corrective to the build-at-all-costs mindset. He realizes that saving humanity requires more than his own ingenuity and that he needs a partner with capabilities he lacks. Companies may not need an alien ally, but they do need a map for the decision.

Before committing money, leaders should:

  • Look three years ahead and include the costs of building, maintaining and improving the capability, rather than just launching it.
  • Compare the last three internal builds with their original estimates for time, cost and staffing.
  • Set an abandonment threshold before spending begins, so management does not keep funding a weak decision.
  • Ask someone without a stake in the outcome to challenge the assumptions.

The practical takeaway is to judge the full operating commitment, not the launch budget.

What do the four decision tests reveal?\n

Strategic importance is necessary, but it is not sufficient. If importance were the only criterion, every business would build its own oil refinery. A worthwhile capability should be difficult for competitors to replicate and easy to integrate into existing operations. Four tests reveal the difference.

  • Defensibility: Does the capability create a durable moat and an advantage that customers notice? If it is easily commoditized and offers only a marginal improvement, building may not justify the investment.
  • Time to value: How quickly does the business need the capability? Building may take too long, while an acquisition can also be time-consuming if an acquisition is required. A closing market window may favor a partner.
  • Integration risk: Every option carries risk. Building transfers ownership risk to compliance, maintenance, staffing and a long tail of fixes. Partnering exposes the company to the provider’s development roadmap and events such as its acquisition. Buying adds hidden technical debt, platform fragmentation and culture clashes.
  • Cultural fit: A buyer must be able to onboard and retain the people required for the long run. A partner presents separate challenges, including managing competitive risks and building trust when the relationship is not exclusive.

No option removes execution risk. Each simply moves it to a different part of the business.

When should a company build, buy or partner?\n

The choices need not be permanent or mutually exclusive. A company may partner first to enter a market and learn, then build or acquire once it understands where the value lies.

The decision rules are direct:

  • Build if: The capability is important, creates a competitive wedge and customer loyalty for the long term, and the company will fund it for years.
  • Buy if: The company needs speed but does not need to own the capability.
  • Partner if: The company needs speed and ownership of the capability.

The framework does not depend on AI. Defensibility, time to value, integration risk and cultural fit apply whether engineers write every line by hand or generate a working demo over a weekend. My conclusion is blunt: “AI changes the cost of starting. It doesn’t yet change the cost of owning.”

The team I argued with could build a streaming device, and eventually did. What it could not build was the position the startup already had. When someone says, “We can build that,” I would thank them for their confidence, let them use AI to build a prototype, and ask to see year three. The next AI-assisted business case should show its year-three operating plan, not only its weekend prototype.

Source: Fast Company

Share this article:

More from Olivia Hart

Olivia Hart

Show full bio

Staff writer covering people, launches and day-to-day practice in entrepreneurship, startups and venture funding at Business Bearings.

2,404 articles

Related articles

« Previous articleNext article »