Lisa Su and AMD

The Professor's Questions

1. What were the strengths of AMD during its rise and into the 1990s and early 2000s?
2. What were the root causes of AMD’s struggles?
3. If you were Lisa Su, taking over AMD in 2014, what would be your first three steps to turn around the company?

My Answers

Q1. What were the strengths of AMD during its rise and into the 1990s and early 2000s?

AMD combined a credible alternative to Intel with real engineering strength and customer value. Its advantage changed as the company grew.

1. A trusted second source. Customers wanted supply options. AMD could enter an established market by providing a reliable alternative. That gave it access to customers and the x86 ecosystem. [S1]
2. Compatibility and value. Customers could use familiar software while gaining another price and performance choice. AMD64 later offered a path to 64-bit computing that protected existing software investments. [S2、S4]
3. The ability to lead in selected technologies. Athlon and Opteron showed that AMD could develop important products of its own. This was more than a low-price position. [S3、S5]
4. Partners that helped turn chips into usable systems. Support from computer makers and software companies mattered because customers buy a working system, not just a processor. [S5]
5. A strong challenger identity. AMD's people-first tradition and engineering ambition could help attract commitment. I would treat this as a cultural asset, not proof that the company always executed well. [S16]

My conclusion: AMD's early opportunity came from offering choice; its stronger position came from making that choice technically valuable.

Q2. What were the root causes of AMD’s struggles?

AMD's struggles came from a mismatch between its ambitions, its resources, and its ability to deliver competitive products consistently. External pressure made that mismatch more costly.

1. An uneven competitive position. Intel had greater scale and strong influence over the PC ecosystem. AMD also faced disputed commercial practices; the 2009 settlement shows that market access was a real issue. But external pressure cannot explain every product and investment decision. [S2、S15]
2. Product choices that did not serve enough customer workloads well. The FX-8150 tests showed a weakness in single-thread performance. My management inference is that a promising design idea must be tested against how customers actually use the product. [S8]
3. Large financial commitments reduced room for error. ATI added valuable graphics capabilities, but its acquisition also used cash and debt. A strategically useful asset can still be acquired with a financial burden that is hard to carry. [S6]
4. The cost and complexity of manufacturing. AMD had to fund design and production while competing at a smaller scale. The foundry separation reduced direct ownership of fabs, but it did not remove supplier, process or delivery risks. [S2、S7]
5. Too much to execute with limited resources. Weak PC conditions, several product lines and a need to rebuild CPU competitiveness competed for attention and funding. My diagnosis is that AMD needed sharper priorities and more reliable execution. This is an interpretation of its business position, not a documented claim that employees were unwilling to cooperate. [S9–S12]

The cycle I would worry about is: weaker products lead to pricing pressure and lost business; lower financial flexibility makes sustained investment harder; further delays then weaken customer confidence. This is a causal framework, not a claim that every link happened in every quarter.

Q3. If you were Lisa Su, taking over AMD in 2014, what would be your first three steps to turn around the company?

I would establish the facts and protect cash, choose a focused product path, and rebuild execution with customers and employees. These steps would overlap; I would not wait for one to finish before starting the next.

First, find out what can keep the company alive and what must be protected.

In the first month, I would review cash needs, debt maturities, inventory, supplier commitments and customer payments with the CFO. I would meet key customers and technical leaders to learn which promises were at risk and which capabilities were still strong. I would build a 13-week cash forecast and a longer product-funding plan. I would stop low-priority spending while protecting critical engineers and customer commitments. The cost is that some projects lose funding and some restructuring may be necessary. An equal percentage cut across every team would be a poor starting point.

Second, put the best people and funding behind a few connected priorities.

I would make high-performance computing the main direction, with a competitive CPU core and graphics as key capabilities. I would maintain useful semi-custom and embedded business to help support the transition, while judging each contract on cash, margin and engineering effort. I would review the CPU work already under way rather than pretend to start it from nothing. Projects outside the main direction would need clear customer evidence, a small budget and a stop rule. The cost is fewer options and greater dependence on a limited number of bets.

Third, make delivery credible again.

I would give each major product one accountable leader, connect engineering, sales and operations in a weekly review, and use shared milestones for performance, power, quality and delivery. I would ask people to report problems early and make it safe to challenge the plan. With key customers, I would agree on realistic validation and launch steps. With the board, I would agree on the funding and evidence needed across more than one product generation. The cost is that problems become visible and some public promises may need to be reduced.

By the end of the first 90 days, I would want three things: a funded survival plan, a clear list of product priorities and cancellations, and a delivery plan owned jointly by the teams and key customers. I would judge early progress by cash visibility, talent retention, engineering milestones and customer validation—not by expecting a new chip to appear in one quarter.

AMD's strength was that it offered a credible alternative to Intel, backed by real engineering. Its weakness was that it could not turn that strength into consistent product and financial performance. If I were Lisa Su in 2014, I would first understand the cash position and protect the people and projects that mattered most. Second, I would focus the roadmap on high-performance computing, while using valuable existing businesses to support the transition. Third, I would rebuild trust through clear ownership, honest reporting and realistic customer commitments. The difficult leadership choice is deciding what to stop, while keeping people committed to a plan whose results will take years to appear.

Case Summary

When AMD first entered the microprocessor market, it held an important position: a credible second source of supply for customers. The Computer History Museum's oral history with Intel insiders records IBM's requirement for a second supplier, and also AMD's reputation at the time as a reliable supplier. That let it enter a market that already had software, PC makers and users, without having to build all the demand on its own. [S1]

In the 1990s, AMD steadily strengthened its own processor design capabilities. After the K6, the Athlon, which began shipping in 1999, became an important product; in 2000 AMD launched an Athlon that reached 1 GHz. In 2003, Opteron brought 64-bit capability to x86 servers while keeping the ability to run existing 32-bit software. This let customers keep their software investments and upgrade their systems gradually. Ecosystem partners such as IBM and Microsoft provided product or software support when Opteron launched. [S2–S5]

AMD also faced bigger rivals, R&D and manufacturing businesses that required constant investment, and pressure to keep iterating its products. When it completed the ATI acquisition in 2006, the announced deal value was about US$5.4 billion, including about US$4.3 billion in cash, with a new US$2.5 billion term loan. The acquisition brought GPU, chipset and other capabilities, but also added financial and integration burdens. Afterwards, in 2009, AMD put its wafer manufacturing business into the newly formed GLOBALFOUNDRIES, gradually moving to a model that relied on outside manufacturing. [S6–S7]

The difficulties on the product side did not end with the manufacturing spin-off. The Bulldozer architecture reached the market in 2011; benchmarks of the FX-8150 at the time showed its single-threaded performance was uncompetitive in some common applications and games. This is evidence about a specific product and specific test scenarios, and can't be written up as "every AMD product lagged at every task." At the same time, weakness in the consumer PC market hurt the traditional business. [S8–S10]

The transformation had already begun before 2014. Jim Keller returned in 2012, reporting to CTO Mark Papermaster and leading CPU core design; Lisa Su joined AMD the same year. In 2013 Sony and Microsoft announced that their next-generation game consoles would use AMD semi-custom chips. In a 2017 interview, Lisa Su said directly that the new high-performance CPU core had been in development for about four years. So the new CPU roadmap, the team and the console business can't all be credited to the period after she became CEO in 2014. [S11–S14]

On October 8, 2014, Lisa Su succeeded Rory Read. Quarterly data from around the handover show a company with divergent business performance: the traditional Computing and Graphics segment was under pressure, while the Enterprise, Embedded and Semi-Custom segment was profitable. What she had to work out was how to get cash, talent, product direction and customer commitments all supporting the same long-term path. [S9–S10, S13]

Key Data

Amounts are kept in their original currency. US$1 billion = 1 billion US dollars; US$1 million = 1 million US dollars.

Dimension	Figure / fact	Source and basis
Early market entry	IBM required a second source; AMD had a reputation as a reliable second supplier	S1, oral history approx. p.5; participants remember details differently, so no single version is invented from it
Own products	Athlon began shipping in 1999; reached 1 GHz in 2000	S2, S3
Ecosystem compatibility	Opteron launched 2003-04-22; AMD64 supports existing 32-bit software as well as 64-bit software	S4, S5; not the same as "the world's first 64-bit processor"
ATI deal	Announced deal value about US$5.4bn; cash about US$4.3bn; new term loan US$2.5bn	S6; announced deal value differs from the accounting purchase price, which includes fees, equity awards, etc.
Manufacturing spin-off	Related transaction completed in 2009; the manufacturing business renamed GLOBALFOUNDRIES	S7; took place before Lisa Su became CEO
Competition dispute	In 2009 Intel settled with AMD, paying US$1.25bn and signing a cross-licence and business-conduct terms	S15; a settlement is not the same as "all allegations confirmed by a final ruling"
Product problems	In 2011 the FX-8150 performed poorly in contemporary single-threaded and some gaming benchmarks	S8; one benchmark is not generalised into a conclusion about all workloads
Talent and projects	Keller returned in 2012; Su joined in 2012; new CPU core development predates the 2014 CEO handover	S11, S13, S14
Customer base	In 2013 Sony PS4 and Microsoft Xbox One announced they would use AMD semi-custom products	S12
CEO handover	2014-10-08, Lisa Su succeeded Rory Read	S13
Liquidity and debt	Cash, cash equivalents and marketable securities at end of Q3 2014: US$938m; total debt about US$2.20bn	S9, S10; the debt has a maturity schedule, so you can't simply subtract to judge bankruptcy or months of survival
Overall performance for the period	Q3 2014 revenue US$1.429bn; GAAP net income US$17m	S9, S10; not "no profit at all that quarter"
Segment divergence	Computing and Graphics revenue US$781m, operating loss US$17m; Enterprise, Embedded and Semi-Custom revenue US$648m, operating income US$108m	S10, segment data approx. pp.26–27; the latter is not a "pure console segment," and operating income is not cash flow
My own calculation	Operating margins of the two segments approx. −2.2% and 16.7%	−17 ÷ 781; 108 ÷ 648. Not gross margins or group net margin
Disclosure timing	The quarter above ended 2014-09-27; results announced 10-16; 10-Q filed 10-30	An operating snapshot from around the handover, not described as public data the market knew before 10-08

Key People

Person / organisation	Role	Function and authority	Relationships and pressures
Jerry Sanders	AMD co-founder, early CEO	Built the early organisation, sales and challenger tradition	Historical influence; not a 2014 decision-maker
Hector Ruiz	CEO at the time of the ATI acquisition	Led strategy and major deals at the time	GPU capability and financial burden became his legacy
Rory Read	Lisa Su's predecessor as CEO	Handed over in 2014, stayed on briefly as an adviser	Carried forward existing transformation work and handed it over
Lisa Su	Joined 2012, CEO from 2014	Primarily responsible for operating priorities, resource allocation and organisational execution	Accountable to the board; works with the technical team, customers and suppliers
Board of directors	Body for appointments and major governance decisions	Supports the CEO, oversees capital and risk	Needs to agree to R&D cycles and financing arrangements
Mark Papermaster, Jim Keller and the engineering team	CTO, head of CPU design, and R&D staff	Technical path and delivery capability	The CEO must provide clear direction, resources and cross-functional support
Intel; NVIDIA	Main competitors in the CPU and GPU markets	Competitive pressure AMD cannot control	Competition on performance, ecosystem, scale and commercial execution
Sony, Microsoft, PC makers; foundry partners	Customers and supply chain partners	Purchasing, design adoption, production collaboration	Customers want delivery credibility; suppliers want visibility on orders and payment

No material was found that is sufficient to identify specific "internal opponents," so no personal conflicts are invented.

Core Tensions

1. Short-term survival and long-term R&D must both hold. Cash is limited, and developing a new core takes years; cutting spending evenly across the board might improve immediate outgoings but damage competitiveness in the next round.
2. There are opportunities in several markets, but only one engineering team. CPU, GPU, consoles, embedded and other directions can't all get equal priority; focusing means giving up some potential revenue.
3. Recognise the value left by predecessors while changing resource allocation. The manufacturing spin-off, the console customers and the CPU team already exist; the new CEO has to judge what to continue and what to scale back.
4. The technology is worth investing in, but customers may not be willing to adopt it right away. Customers still need to validate software, quality, supply and the future roadmap; R&D and commercial trust have to be built together.
5. A long-term direction needs patience; project management needs timely course correction. Sticking with a direction can't become an excuse for tolerating hidden problems or endless delays.