Read on Twitter
THREAD AWS $AMZN
AMZN has 15Xed in 6 years (~$60bn->~$1tr). Along the way there have been many investor concerns, which have in hindsight been "can't see the forest for the trees" nitpicks: 1) AWS not profitable; 2) AWS price cuts; 3) Depr acctg; 4) Too much start up exposure.
AMZN has 15Xed in 6 years (~$60bn->~$1tr). Along the way there have been many investor concerns, which have in hindsight been "can't see the forest for the trees" nitpicks: 1) AWS not profitable; 2) AWS price cuts; 3) Depr acctg; 4) Too much start up exposure.
First to establish valuation. I don’t think it is controversial to say that AWS is worth ~$1tr+ today annualizing nearly $50bn in revs this past Q, ~35% GAAP Op Margins growing ~30% in COVID times. In 2021, AWS will likely put up around ~$58BN in revenue (MSe).
While IaaS is not SaaS, applying even the lower end of Software universe multiples to this level of growth (not to mention profitability or any analysis of how great the business is) would get you to a $1trillion value: 16X EV/Sales ~$60bn in revs ~= $1tr.
And AWS is a lot more profitable than most software companies (today). On a P/E basis, this equates to ~60X 2021 GAAP EPS (assumes 20% Tax rate).
Looking back on it, it is funny to think about a few of the past AWS bear cases/worries:
Looking back on it, it is funny to think about a few of the past AWS bear cases/worries:
1) AWS not profitable
AMZN first disclosed AWS financials on the 1Q15 print. Before AMZN disclosed AWS revs/EBIT, the consensus was the MORE unprofitable AWS was, the better it was for AMZN’s stock because it meant the retail business was even more profitable than people thought.
AMZN first disclosed AWS financials on the 1Q15 print. Before AMZN disclosed AWS revs/EBIT, the consensus was the MORE unprofitable AWS was, the better it was for AMZN’s stock because it meant the retail business was even more profitable than people thought.
"If AWS had double digit negative margins, AMZN retail could even be more profitable than Walmart (OPM%, not gross $)!" - one sell side report
On April 23, 2015 AMZN reported 1Q15 results breaking out AWS revs/profits for the first time. AWS reported $1.5bn in revs growing ~50% y/y (accelerating to 80+% next Q). Shockingly AWS was profitable. Very profitable. 17% Adj Op Margins with 50% adj. EBITDA margins!
And investors loved it as the stock was up nearly 15% the next day or $26bn (strong retail results helped). Even with the surprisingly good numbers/profitability, one bank (MS) only put AWS value at $34bn, or 10X the just reported quarterly EBITDA run rate. JPM $60bn. GS $85bn.
Using $60bn for 2015 valuation, If AWS is valued at $1tr today, this means that 6 years later, AWS’s value 15Xed, or compounded 60%. What does this mean for AMN Retail? It implies retail biz was valued at ~$130bn and today is worth ~$600bn, 360% return, or 29% annualized.
2) AWS Price Cuts. In 2015-2017, I remember tracking the number of AWS price cuts and sending various tables to clients trying to estimate the impact (AWS announces them on their blog https://aws.amazon.com/blogs/aws/category/price-reduction/).
How much will price cuts impact growth this year? What about margins? It did cause lumpiness in the earlier days (1Q15s 50% growth rate accelerated to 80+% in 2Q after lapping a price cut among other things).
It also created a barrier to entrants. We have the Big 3 today with AMZN still bigger than the #2-5 combined, and it aided AMZN's moat as it built higher value add (and margin) products on top of S3/EC2 (RedShift, DynamoDB, etc). Try to find Price Cuts impact to growth here.
3) AWS accounting shenanigans with depreciation
After AMZN disclosed the AWS metrics, valuation hawks jumped all over the AWS disclosures. 50% adj. EBITDA margin were masked by 1) capital leases and 2) depreciation shenanigans.
After AMZN disclosed the AWS metrics, valuation hawks jumped all over the AWS disclosures. 50% adj. EBITDA margin were masked by 1) capital leases and 2) depreciation shenanigans.
Servers were being capitalized and depreciated over 3 years because that’s how the GAAP accounting rules accounted for typical server and computer equipment but Public Cloud was different given the higher utilization levels and actual useful lives were actually likely <3 years.
A 2 year useful life would increase depr costs by 50% making Op margins effectively 0%. In late 2019, AMZN announced that due to better server useful life than they had previously though, it would extend the depr schedules from 3 years to 4 years. MSFT soon followed suit.
Does this matter? Not for cash flow (in the ST, LT yes as capex should be more efficient). But AMZN’s GAAP OI has improved ~500bps this year to 35% largely from this accounting change. More importantly, this change proved the 2 year useful life theory wrong.
4) Too much start-up exposure, how would AWS handle the 2015/16 tech downturn
I remember the 2015/16 tech downturn vividly. I was on the sell side covering smidcap internet. Almost half of my smidcap Internet names down 40+% earnings in November 2015.
I remember the 2015/16 tech downturn vividly. I was on the sell side covering smidcap internet. Almost half of my smidcap Internet names down 40+% earnings in November 2015.
People were saying this was the tech Bubble all over against. PrivateCo valuations were poo pooed. I questioned if smidcap Internet was even worth covering. And for AWS, investors questioned how much this downturn would hurt.
This was when AWS was seen as renting storage and compute for cheap for those that couldn’t afford servers of their own. Enterprise adoption was very early days.
Having fast growing, innovative companies was seen as a WEAKNESS rather than a customer base that would grow and be future leaders (nice customer base to have if you can get it). A few quarters later, smidcap internet/software rebounded and numbers started looking good again.
Conclusion? If you have a great business growing fast with strong unit economics that you think COULD be the future of something big (Public Cloud), Don't Sweat the Small Stuff.
