STRAT 390 - Readings

[Themes] Which case is about corporate diversification?

Amgen: should an innovator biotech also enter biosimilars (imitation)?

[Themes] Which case is about vertical integration?

Delta: the airline bought its own oil refinery in 2012 to control jet-fuel costs.

[Themes] Which case is about transaction costs and the scope of the firm?

Nvidia: should it keep integrating into systems/cloud (competing with customers) or stay a chip supplier, and how should it handle its dependence on TSMC?

[Themes] Which case is about global expansion and globalization?

NFL: why it lagged the NBA, MLB and NHL internationally, and where, how and who should lead expansion abroad.

[Themes] Which case is about localization on a global scale?

Nestlé KitKat Japan: a global brand made locally meaningful as an exam good-luck charm ('kitto katsu').

[Themes] Which case is about organizing for corporate advantage (central vs. local decisions)?

Pernod Ricard: a decentralized company centralizes the building of its digital tools (D-STAR, Matrix) at HQ while markets handle execution.

[Themes] Which case is about M&A decisions?

Microsoft-Activision Blizzard: a $68.7B deal for gaming content, cloud gaming, mobile and the metaverse; did the price and rationale make sense?

[Themes] Which case is about acquisition strategy for multi-sided platforms?

Uber-Cornershop: Uber bought 51% of a Chilean grocery-delivery platform and kept it independent.

[Amgen] Core decision (Sept 2009)

CEO Kevin Sharer had to decide whether Amgen, the world's largest biotech innovator, should also develop and sell biosimilars ('me-too' biologics). Moving from pure innovation into imitation is a diversification move, and he had to recommend it at the October Board strategy review.

[Amgen] Sharer's three questions

(1) Would biosimilars be a good business to be in (is the industry attractive)? (2) Could Amgen do well in it (can we win)? (3) How would entering affect the Amgen organization?

[Amgen] Biologics vs. small molecules

Biologics are large, complex molecules (about 20,000 atoms), such as proteins and antibodies. They are made in living cells using recombinant DNA and are mostly injected in hospitals. Small molecules have about 80 atoms and are made by predictable chemical synthesis (the 'pill industry').

[Amgen] Biosimilar

A copy of a biologic sold after the original's patent expires. It can't be identical because the copier uses a different cell line and process, so it is 'similar' rather than generic. It needs some clinical trials and careful manufacturing.

[Amgen] Generics (contrast to biosimilars)

Exact copies of small-molecule drugs. Under the Hatch-Waxman Act (1984) they only need to show bioequivalence, through lab tests. Development costs about $1M-$10M, pharmacies can substitute them automatically, and they quickly take most of the market (Coreg: 80% of prescriptions within months, at about 20% of the original price).

[Amgen] Why biosimilars are harder than generics

Molecules are complex and hard to copy exactly (about 250 in-process tests vs. about 50 for small molecules). Some clinical trials are needed. Regulators decide approval case by case. Europe did not allow automatic substitution, and post-launch pharmacovigilance (safety monitoring) is required.

[Amgen] Arguments FOR entering biosimilars

Amgen could reuse its strengths: process development, top-quality manufacturing, regulatory relationships, and selling to hospital clinicians. It had excess manufacturing capacity. Many rivals' blockbusters (Avastin, Herceptin, Rituxan, Lucentis) would lose patents by 2019, and Amgen needed new growth.

[Amgen] Arguments AGAINST entering biosimilars

Amgen would lose credibility after years of arguing biosimilars were unsafe. Its innovator culture ('we gold plate everything') clashes with being a low-cost imitator. Many entrants would push prices down, rivals would learn to make biosimilars over time, and some targets would compete with Amgen's own drugs (Enbrel vs. Humira/Remicade; Vectibix vs. Erbitux).

[Amgen] Bradway's business-case setup

He treated biosimilars as a separate business that could use some Amgen resources, assumed US rules would be similar to Europe's, and analyzed only the US and Europe. Scenario A (lower competition): 2 biosimilars, 40% volume penetration, prices at 70% of the original. Scenario B: 5 biosimilars, 60% penetration, prices at 50%.

[Amgen] Course concept: Porter's diversification tests

Attractiveness test (is the industry structurally profitable?), cost-of-entry test (does the cost of getting in eat up the profits?), better-off test (does combining with the corporation create a competitive advantage?)

[Amgen] Course concept: related diversification and scope

Value comes from sharing resources and capabilities across businesses (for Amgen: manufacturing, regulatory skill, clinician relationships). The risk is negative synergy when two businesses need different cultures and strategies (innovator vs. imitator).

[Amgen] Risk of staying out

Generics firms (e.g., Sandoz), big pharma (Merck BioVentures, which bought Insmed's biosimilar portfolio for $130M) and Asian players (Samsung, LG Life Sciences) were entering. Amgen's own drugs would face biosimilars either way.

[Delta] What Delta did

In 2012 Delta paid $150M for a ConocoPhillips refinery near Philadelphia, which ConocoPhillips had been shutting down. This is backward vertical integration into jet-fuel supply.

[Delta] Why Delta bought the refinery

Volatile fuel prices helped push Delta into bankruptcy in 2005, and it later lost money hedging with financial instruments. The crack spread was its fastest-growing cost. CEO Richard Anderson: 'We were not going to be controlled by fuel prices.'

[Delta] Crack spread

The gap between the price of jet fuel and the price of crude oil, i.e., the refiner's margin. Owning the refinery lets Delta keep this margin instead of paying it to outside refiners.

[Delta] How the refinery is structured

A subsidiary runs it and sells jet fuel to Delta at market rates, so Delta stops paying outside suppliers for refining.

[Delta] Benefits of the refinery

Operating profit in most years, protection from Northeast supply disruptions, and jet fuel cheaper by several cents a gallon. It saved about $785M in 2022 (Ukraine war) and was expected to add $300M to Q2 2026 earnings during the Iran war, offsetting about 1/3 of the crack-spread increase.

[Delta] Drawbacks of the refinery

Lost more than $200M in 2020 when Covid crushed fuel demand. The plant is over 100 years old and has needed $1.6B in capital spending. Other costs: ethanol-blending compliance credits, environmental liabilities and operational risk, and continued exposure to crude oil prices.

[Delta] Supporter of vertical integration

Economist Philip Verleger: a smart move, because oil companies treated airlines as cash cows. United CEO Scott Kirby: over the refinery's life everyone shares the extra supply, but at today's high crack spread the benefit is unique to Delta.

[Delta] Why airlines stopped hedging

Airlines' pricing power let them pass higher fuel costs on to fliers without paying for expensive insurance

[Delta] 2026 context

After US/Israeli strikes on Iran, jet fuel roughly doubled to about $4.80/gal and the crack spread hit a record. Airlines faced about $40B in extra fuel costs, and Delta expected its Q2 fuel bill to rise $2B. It raised bag fees and trimmed flights.

[Delta] Course concept: backward vertical integration

Moving into a supplier's stage of the value chain (make instead of buy). Pros: secure supply, capture the margin, less exposure to supplier power, better coordination. Cons: capital needs, lack of expertise, less flexibility, and exposure to the upstream industry's cycles.

[Nvidia] Huang's two decisions (fall 2024)

(1) How to allocate and price scarce GPUs. (2) Where to play in the value chain: keep integrating into full systems and cloud (competing with customers) or stay a component supplier, and how to hedge its dependence on TSMC.

[Nvidia] Fabless model

Nvidia designs GPUs but outsources fabrication to TSMC, assembly and testing to third parties, and boards to contract manufacturers. Intel and Samsung are vertically integrated. Outsourcing avoids roughly $25B per fab but gives Nvidia less control of supply.

[Nvidia] Dependence on TSMC

TSMC makes over 90% of leading-edge logic chips and can demand prepayments. Blackwell was delayed by poor TSMC yields. Risks: Taiwan geopolitics and 18-24 months to build a new fab. The alternative, Intel's foundry, is behind and is also a competitor.

[Nvidia] ARM deal (attempted)

A $40B deal (2020), abandoned in 2022 after UK and FTC antitrust objections. Nvidia then built its own ARM-based CPU (Grace).

[Nvidia] DGX H100

Nvidia's complete AI server (up to 8 H100s, up to about $500K). This forward integration put Nvidia in competition with OEM customers Dell, HP and Lenovo.

[Nvidia] DGX Cloud (2023)

Nvidia rents servers inside Google, Microsoft and Amazon data centers and re-rents them to end customers (e.g., ServiceNow, Amgen), competing with the hyperscalers who are its biggest customers.

[Nvidia] Customer concentration

About 98% share of data-center GPUs. Data centers were roughly 80-90% of revenue, and nearly half of revenue came from 4 large data-center customers.

[Nvidia] Hyperscalers integrating backward

Google TPU (about $2-3K per chip vs. about $15K for an Nvidia GPU), Amazon Trainium/Inferentia/Graviton (from its Annapurna Labs acquisition), and Microsoft all build in-house chips to cut the 'Nvidia tax' and avoid shortages.

[Nvidia] Direct competitors

AMD (cheaper GPUs; bought ZT Systems for $4.9B to build custom systems; ROCm software), Intel (Gaudi chips from its $2B Habana Labs purchase), and startups (Cerebras, Groq, D-Matrix, Tenstorrent).

[Nvidia] GPU allocation options

An auction (highest prices, concentrated customers), price discrimination by willingness to pay, strategic allocation (e.g., 40K GPUs to CoreWeave), or a queue that favors customers who buy other Nvidia products. Any choice has to survive the DOJ antitrust probe (Aug-Sep 2024).

[Nvidia] Demand risks

Semiconductor boom-bust cycles and over-ordering, data centers overbuilt like the dark fiber of 2000, smaller LLMs running on devices (Llama, Apple Intelligence), inference moving to cheaper chips, China export controls (14% of data-center sales in FY24), and power limits.

[Nvidia] Course concept: transaction cost economics

A firm's boundaries come from comparing the cost of using the market (searching, negotiating, monitoring, hold-up) with the cost of doing it in-house. Integration makes sense when asset specificity, uncertainty and frequency are high and hold-up risk is serious.

[Nvidia] Hold-up in the Nvidia case

Nvidia risks hold-up by TSMC (sole supplier with pricing power). Customers risk hold-up by Nvidia (near-monopoly GPU supplier), so they integrate backward into their own chips.

[Nvidia] Course concept: downsides of expanding scope

Competing with your customers (channel conflict) can push them to rivals. Integration also adds capital needs and reduces focus and flexibility. Staying a component supplier keeps customers as partners.

[NFL] Core question

Why had the NFL failed to globalize like the NBA, MLB and NHL? Should it try again, and if so where, how, and who should lead: the league or individual teams?

[NFL] Revenue sharing

The NFL splits national and local TV money equally among its 32 teams, and gate receipts 60% home / 40% league. This creates parity ('Any given Sunday') but weakens any single team's incentive to invest abroad. The NHL has no revenue sharing, which gives it a stronger reason to go overseas.

[NFL] NFL's international revenue gap

Only 2% of NFL media revenue comes from outside the US (NHL 42%, NBA 21%, MLB 7%), even though the NFL is the most valuable league ($18.6B revenue in 2022).

[NFL] American Bowl (1986-2005)

40 preseason exhibition games abroad. Mostly backups played and teams didn't care about the result, so the games lacked competitive energy.

[NFL] World League / NFL Europe / NFL Europa (1991-2007)

A separate developmental league abroad. It lost about $20M a year and produced few stars (Kurt Warner, Jake Delhomme); by the end 5 of 6 teams were in Germany. Lesson from Mark Waller: 'It's got to be the NFL. It can't be the European League.'

[NFL] International Series (2005-present)

Regular-season games abroad: Mexico City in 2005, London's Wembley in 2007, 3 UK and 2 Germany games in 2023, and Brazil (São Paulo) in 2024. The top-down logic: build fans, then media and sponsorship, and maybe eventually a franchise.

[NFL] Why games abroad are hard to scale

Only 17 games a season and about 9 home games per team; owners like Jerry Jones don't want to give up home games; travel fatigue creates a competitive disadvantage.

[NFL] Fan and product obstacles

Complex rules, 3.2-hour games with constant stoppages, and loyalty built on local 'my town vs. your town' rivalries. The Super Bowl is a one-day spectacle, 'like a Broadway play': you go once and go home.

[NFL] Talent obstacles

Nearly all players come from the NCAA. Football needs infrastructure, coaching and full teams, unlike basketball or soccer on a playground. The NFL tried an International Development Practice Squad Program (2004).

[NFL] Marketing obstacles

Average career is 3.3 years (running backs 2.6), and helmets and pads hide players, so global stars are hard to build. New openings: star QBs (Mahomes' $450M+ deal) and Travis Kelce / Taylor Swift.

[NFL] Global Markets Program

Individual teams get marketing rights in specific foreign markets (21 clubs in 14 markets in 2023) to build local fan loyalty.

[NFL] Flag football

Low barriers to entry (little equipment, simpler rules). It will be in the 2028 LA Olympics, a possible 'Dream Team' moment, and it has boosted female participation.

[NFL] Samoa lesson

With only about 45,000 people, Samoans make up about 3% of NFL players, thanks to missionaries and US-funded high schools with football programs. Grassroots infrastructure creates talent.

[NFL] Where to expand?

Mexico and Canada are easiest (existing interest, proximity) but small next to the EU and Asia. Germany, the UK and Brazil (about 20M fans) are growing targets.

[NFL] Who should lead abroad?

The league, which preserves revenue sharing and parity, or individual teams, which have stronger incentives but could erode the revenue sharing that makes the NFL great.

[NFL] Course concept: CAGE distance

Cultural, Administrative, Geographic and Economic distance between home and foreign markets. The NFL faces high cultural distance (the sport is unfamiliar), geographic distance (travel) and economic distance (willingness to pay).

[KitKat Japan] Core decision (2008)

Masafumi Ishibashi, Nestlé Japan's confectionery marketing manager, had to figure out how to keep deepening KitKat's place in Japanese culture as rivals copied its success.