11 · Who captures the value

The canonical case of the framework, discussed by Teece himself, is computed axial tomography. The CT scanner is developed in the laboratories of the British company EMI by Godfrey Hounsfield, who will receive the Nobel Prize in medicine for it in 1979: an authentic radical innovation, the greatest advance in diagnostic imaging since X-rays, launched commercially in 1972 with years of advantage over anyone else.

Eight years later, EMI was out of the market it had created, its business sold to a competitor, and the CT market belonged to General Electric and Siemens. The reconstruction with the three variables explains the outcome without any need for spectacular errors. Appropriability was weak: patents existed but did not cover the essentials in a blocking way, and the architecture of the machine, once seen, taught a great deal to the giants of electromedical equipment, which had the engineering capacity to reinvent around the claims. The dominant design was not yet fixed: generations of scanners followed one another rapidly, and the pioneer's advantage was wiped out at each turn of architecture. And above all the specialised complementary assets were all on the other side: GE and Siemens had hospital sales networks built over decades, widespread technical support (decisive for a machine on which operating lists depend), reputation with radiology departments, production and regulatory capacity; EMI, which came from consumer electronics and music, had to build all this from scratch, in a market, the United States hospital market, where its competitors were at home. The better technology with a weak regime and other people's assets: the value migrated exactly where the framework predicts, and the Nobel remained with those who had created the value that others captured.

The case fixes the methodological lesson: the question "is the technology superior?" is not enough to found an investment; the complete question is "given these three variables, does the share of value that this company can capture justify the capital?". Sometimes the honest answer is that the value is there but will go to others, and that the right strategy is not to compete but to sell or license to whoever owns the assets: which is the bridge to section 4.

There is a third way beyond competing and selling: keeping ownership of the technology and transferring its use. The licensing model turns the logic of complementary assets on its head: instead of building what is needed to reach the market, you enlist as licensees those who already have it, and you capture value as a fraction of theirs, multiplied by all of them.

The case that shows the scale of this is ARM. The Cambridge company designs processor architectures and manufactures nothing: it licenses the designs to those who produce (practically the whole semiconductor industry for mobile devices), collecting royalties on every chip. The model has turned what would have been an impossible position (competing in manufacturing against giants with foundries costing tens of billions) into the opposite one: the giants are customers, the complementary assets are theirs, and appropriability rests on a portfolio of intellectual property and on architectural know-how that the licensees have an interest in seeing prosper, not in eroding. The conditions for the model to work must, however, be declared: a strong appropriability regime is needed (without defensible exclusivity, the licensee becomes an imitator), a technology separable from the final product (a component, an architecture, a method: not every innovation is), and the capacity to serve the licensees over time, because a licence is a relationship, not a sale.

Licensing brings with it an architectural principle that holds outside its business model too: the separation between ownership of the intellectual property and the vehicle that uses it. When the IP is held by a party (or by several owners) distinct from the operating company, which uses it under licence, three effects are obtained. The first is asset protection: if the operating vehicle fails, the IP is not among its assets in the insolvency; it survives the accident, in the hands of those who created and financed it, and remains available for a new initiative, a sale or a licence to third parties (it is the residual value of stopped projects, chapter 08). The second is reusability: the same base technology can serve several vehicles, several markets, several attempts, without the fate of one dragging it down. The third is clarity of incentives: whoever created the knowledge remains its owner and is remunerated by its use, which aligns the protection of the investor with the recognition of the creator. The price of the architecture is its contractual complexity (well written licences, exclusivities and fields of use defined, the fate of the licence in pathological cases), which chapter 12 will treat at the level of instruments.

• D.J. Teece, "Profiting from Technological Innovation" (Research Policy, 1986): the founding article; thirty-five pages that are worth reading in full, EMI case included.
• W.M. Cohen, R.R. Nelson and J.P. Walsh, "Protecting Their Intellectual Assets" (NBER Working Paper 7552, 2000): the reference survey on what really protects, sector by sector.
• D.J. Teece, "Business Models, Business Strategy and Innovation" (Long Range Planning, 2010): the completion of the framework on the side of the business model as a mechanism of capture.
• On the ARM model: the public company documentation (annual reports and the listing prospectus) is the most instructive primary source on the economics of licensing at scale.