AMD FY23 result - the tiger by the tail
ADVANCED MICRO DEVICES –The Tiger by the Tail
Broadly my international investing strategy is to accumulate
quality stocks at reasonable prices. These will be financed by an existing exposure
to an international ETF. The companies I would like to acquire would be leaders
in their industry.
Where does AMD fit into this strategy? AMD is not the leader
in semiconductor design, that company is NVDA, especially in GPU’s, maybe AMD
is a contender in CPU’s. The truth is I stuffed up NVDA by selling the stock after
I entered at a reasonable price, attempting to time the market. NVDA is now
much higher, I am left with AMD.
That said AMD is no slouch when it comes to eps growth. Analysis
of AMD is a little complex. In the US the difference between GAAP and non-GAAP
earnings is mainly due to SBC (share based comp) and amortisation of goodwill. Interestingly,
one is a real expense (SBC) and the other is not, both are significant numbers
for AMD making the analysis a bit complex. The other issue is that AMD bought Xilinx
through an all-share deal which impacts ROE and comparable numbers.
The semiconductor design industry is influenced by cyclical
issues, secular growth and an ongoing product cycle innovation, that makes earnings
prediction quite tricky. The overriding new opportunity in the industry is the
use of mainly GPUs but also CPUs in the advancement of LLM in training and
inferencing for AI. The major beneficiary of this new and huge demand has been
NVDA. AMD is playing catch-up and is unlikely to close the gap any time soon,
but the demand could be so strong that the second player may do quite well.
Estimating the amount of demand years into the future is a
challenge. AMD has estimated a TAM of $400B for accelerator chips, mainly GPU,
ASIC and memory by 2027. That’s a big number when AMD's current revenue is $22b.
A 10% share would be a doubling of revenues. Given customer preference for not
dealing with a monopoly, AMD has the potential to participate in mainstream
growth and particularly niche markets.
AMD has four divisions, Client, Gaming, Data centres and Embedded
(basically XLNX). Gaming looks cyclical, Client (PC) is cyclical but may further
grow through AI-enhanced PCs. Embedded should see growth through increases in
edge computing over time. The huge growth will come from DCs. AMD is now prepared
to participate in this growth.
AMD's balance sheet is strong with net cash and positive FCF
but can be strained in cyclical downturns due to ongoing R&D requirements.
Valuation, as can be imagined, is tricky. AMD reported about
$2 EPS adjusting for the two accounting issues mentioned above. At $160 that is
80X historic earnings. If AMD reach approximately 10% share by 2028 that is 31%
EPS growth for 5 years. At an exit PE of 20X gives a return of 5% pa at $163.
Given the volatility likely in demand over the next few years that is not a great
trade-off.
Under different valuation scenarios, we can get a likely
sell target of around $250, and around $100 as a buy. The large range is due to
the considerable uncertainties. Modest allocation to a likely strong and
volatile growth company.
Comments
Post a Comment