Great set of thoughts. Should definitely consider a part 2 on how to play this. It’s becoming more crticial to visualize the future yet vastly more difficult to do so.
The chip biz has lived this since it's inception. Been in the chip biz since '76, done many generations of product planning. "Why" and "Who needs that" are always questions. but over time, the right answer was 'you don't have to be smart to do product planning in the semiconductor biz, you just need semi-log paper".
How quaint.. who knows what semi-log paper is anymore ;-)
One CTO I worked w/ had a great phrase. You don't shoot ahead of the duck. You shoot ahead of next year's duck. That CTO is still at it, driving one of the industry's leaders.
Loved the article, exponential growth impacts not just one industry....
Great article! One caveat I wonder about though, for all of recent human history population growth has been exponential but that seems to have stopped and we will see peak population and decline (already starting to see in the west caveat immigration). How much growth has been driven by population growth till now? TAM has been increasing but perhaps no longer.
I’m surprised how you just brushed over the threat of climate change - closing your eyes and trusting that tech will take care of the problem may not work for the climate crisis in the way it has for other issues. I suggest reading “The Uninhabitable Earth.”
1) Exponential growth: its always fascinating to see a parabola, but ultimately exp is roughly doubling, meaning at any point in the graph the users sees a steep line ahead
2) Valuation: I guess in light of the curve-discussion, the question is also: how much forward-looking exponent is already priced in. Obviously, long-term everything that is not bankrupt grows (thanks to devaluation), but mid-term is more tricky. After the spring and summer comes the winter. Living through decade of sideways movement is deranging to most investors
Don’t flame me but — curious to know where the value is at Uber? Even darling Amazon has extremely shady accounting practices that allow it to subsidize it’s insane revenue growth… what if part of this insanity is that it is actually insanity like the south sea bubble?
If you use Uber and enjoy the service, the value is how much $$ you pay per ride. That's consumption that would have accrued elsewhere in the past (public transport, drive yourself or a taxi) or consumption that could never have happened. Uber expanded the market and drove consumption for ride hailing.
Amazon - same thing. Think about the countless billion dollar businesses running on AWS (MongoDB, Snowflake etc) and the millions of people buying stuff on Amazon. That has tangible value.
If you're talking about "shady accounting practices", let's start by defining what that shadiness is. Could you elaborate more. I read both Uber and Amazon form 10ks and could not find anything that is considered "shady" other than usual things like capitalising expenses that generate meaningful returns and using that to decrease tax. I'm okay with that since that generates value long term through their high return on capital.
I thought economic value had something to do with profits. Apparently negative unit economics don’t matter at scale? Uber loses money on every ride, Amazon runs AWS at an economic loss etc.. Strange concept of value.
I can't justify investing in Uber because I can't see operating leverage at the moment. But if you define Uber's value to yourself as a user / customer, it's pretty obvious.
As for AWS, it made $45bn (Jan 2021 Form 10K). They added $16bn worth of infrastructure to AWS and depreciated $7bn. I'm not sure why you think AWS is being run at an economic loss. Perhaps instead of sarcastic jibes like "Strange concept of value", put in more effort into your arguments by referencing financial statements and pull out data that can convince me that AWS is running at an economic loss.
AWS is built with capital leases (on which you get to take a depreciation charge). This is in essence debt. It’s a shell game they can keep playing but not forever because this debt has to be serviced and paid back. We know Amazon loses money on its retail business, subsidizing shipping. And much of the cash it generates there isn’t amazon’s to spend but needs to be paid to sellers. This negative unit economic at gigantic scale.
Side question: how does google lose 4bn with its cloud computing and amzn is a cash machine?
You're right in pointing out the massive capex. But you're incorrect in saying it is a shell game they can't play forever. Debt is very useful for a large company like Amazon especially if the cost of that debt is low relative to the return on invested capital. They spend 16b on AWS infrastructure and it is growing at 30+% a year. That 16b goes towards a business unit that will become $58b next year. Operating income actually grew 37% to 3.5b on that 45b revenue. This shows that AWS has operating leverage. Consumption of their infrastructure outstrips the operating costs. As long as they can keep financing AWS infrastructure growth at reasonable debt levels, it will continue compounding and increasing the business' overall free cash flow.
Google Cloud is a younger business that has yet to clear the hurdle of operating expenses. But it will eventually as it scales.
IaasS (infrastructure as a service) is an incredibly capex heavy business and the reason few companies do it (Microsoft, Amazon, Google, Tencent, Alibaba, Huawei, JD) is that these businesses have other businesses that kick up a lot of free cash flow that can service the debt required to invest in IaaS.
Going back to Amazon, it has total outstanding debt larger than cash and receivables on the balance sheet. I'd be worried if Amazon didn't have business units that generate cash flow that's more than enough to service the debt obligations + is able to invest that debt in high return ventures like AWS, Amazon logistics and core commerce.
From a pure balance sheet and valuation perspective, I prefer Alibaba at the moment by the way. It's got a better balance sheet (cash to debt ratio), but this is normal since Alibaba is far more asset light than Amazon.
Another business you can look at with similar debt loads is Costco. The balance sheet has 28b in cash and about 24b in current debt. But Costco runs a capex heavy business. It has to spend a lot of money upfront building out new warehouses which then mature over a 5-7 year period. They are able to fund this from compounding cash flow in each store.
Basically my point is : Debt is necessary and good as long as you can fund it easily with growing cash flow + your own Free Cash Flow continues growing after paying the monthly debt repayments. This means your ROIC is higher than debt obligations.
You make very salient points. I’m clearly in the minority here. What I’m saying is that what they describe as FCF isn’t actually FCF. It’s money they either owe merchants (this isn’t their cash, it’s a very temporary float plus subsidized shipping cost) or, with AWS, selling more and more server space on servers that are never paid for (AWS is also becoming a commodity fast!).
Interesting perspective on S-curves within a backdrop of exponential growth.
I think your analysis on trend extrapolation was great. Reminded me of one of the concepts explored in "Factfulness".
interesting..thanks for sharing!
Great set of thoughts. Should definitely consider a part 2 on how to play this. It’s becoming more crticial to visualize the future yet vastly more difficult to do so.
The chip biz has lived this since it's inception. Been in the chip biz since '76, done many generations of product planning. "Why" and "Who needs that" are always questions. but over time, the right answer was 'you don't have to be smart to do product planning in the semiconductor biz, you just need semi-log paper".
How quaint.. who knows what semi-log paper is anymore ;-)
One CTO I worked w/ had a great phrase. You don't shoot ahead of the duck. You shoot ahead of next year's duck. That CTO is still at it, driving one of the industry's leaders.
Loved the article, exponential growth impacts not just one industry....
Great read, especially in 2022. It'd be awesome to see a reflection on this piece given all the upheaval and change we've seen this year.
Unfortunately, the same expo acceleration seems to be happening to the climate as well.
Great article! One caveat I wonder about though, for all of recent human history population growth has been exponential but that seems to have stopped and we will see peak population and decline (already starting to see in the west caveat immigration). How much growth has been driven by population growth till now? TAM has been increasing but perhaps no longer.
What happened to ur article abt the cosmos ecosystem?
I’m surprised how you just brushed over the threat of climate change - closing your eyes and trusting that tech will take care of the problem may not work for the climate crisis in the way it has for other issues. I suggest reading “The Uninhabitable Earth.”
Hmm, my $0.02
1) Exponential growth: its always fascinating to see a parabola, but ultimately exp is roughly doubling, meaning at any point in the graph the users sees a steep line ahead
2) Valuation: I guess in light of the curve-discussion, the question is also: how much forward-looking exponent is already priced in. Obviously, long-term everything that is not bankrupt grows (thanks to devaluation), but mid-term is more tricky. After the spring and summer comes the winter. Living through decade of sideways movement is deranging to most investors
Don’t flame me but — curious to know where the value is at Uber? Even darling Amazon has extremely shady accounting practices that allow it to subsidize it’s insane revenue growth… what if part of this insanity is that it is actually insanity like the south sea bubble?
If you use Uber and enjoy the service, the value is how much $$ you pay per ride. That's consumption that would have accrued elsewhere in the past (public transport, drive yourself or a taxi) or consumption that could never have happened. Uber expanded the market and drove consumption for ride hailing.
Amazon - same thing. Think about the countless billion dollar businesses running on AWS (MongoDB, Snowflake etc) and the millions of people buying stuff on Amazon. That has tangible value.
If you're talking about "shady accounting practices", let's start by defining what that shadiness is. Could you elaborate more. I read both Uber and Amazon form 10ks and could not find anything that is considered "shady" other than usual things like capitalising expenses that generate meaningful returns and using that to decrease tax. I'm okay with that since that generates value long term through their high return on capital.
I thought economic value had something to do with profits. Apparently negative unit economics don’t matter at scale? Uber loses money on every ride, Amazon runs AWS at an economic loss etc.. Strange concept of value.
I can't justify investing in Uber because I can't see operating leverage at the moment. But if you define Uber's value to yourself as a user / customer, it's pretty obvious.
As for AWS, it made $45bn (Jan 2021 Form 10K). They added $16bn worth of infrastructure to AWS and depreciated $7bn. I'm not sure why you think AWS is being run at an economic loss. Perhaps instead of sarcastic jibes like "Strange concept of value", put in more effort into your arguments by referencing financial statements and pull out data that can convince me that AWS is running at an economic loss.
AWS is built with capital leases (on which you get to take a depreciation charge). This is in essence debt. It’s a shell game they can keep playing but not forever because this debt has to be serviced and paid back. We know Amazon loses money on its retail business, subsidizing shipping. And much of the cash it generates there isn’t amazon’s to spend but needs to be paid to sellers. This negative unit economic at gigantic scale.
Side question: how does google lose 4bn with its cloud computing and amzn is a cash machine?
You're right in pointing out the massive capex. But you're incorrect in saying it is a shell game they can't play forever. Debt is very useful for a large company like Amazon especially if the cost of that debt is low relative to the return on invested capital. They spend 16b on AWS infrastructure and it is growing at 30+% a year. That 16b goes towards a business unit that will become $58b next year. Operating income actually grew 37% to 3.5b on that 45b revenue. This shows that AWS has operating leverage. Consumption of their infrastructure outstrips the operating costs. As long as they can keep financing AWS infrastructure growth at reasonable debt levels, it will continue compounding and increasing the business' overall free cash flow.
Google Cloud is a younger business that has yet to clear the hurdle of operating expenses. But it will eventually as it scales.
IaasS (infrastructure as a service) is an incredibly capex heavy business and the reason few companies do it (Microsoft, Amazon, Google, Tencent, Alibaba, Huawei, JD) is that these businesses have other businesses that kick up a lot of free cash flow that can service the debt required to invest in IaaS.
Going back to Amazon, it has total outstanding debt larger than cash and receivables on the balance sheet. I'd be worried if Amazon didn't have business units that generate cash flow that's more than enough to service the debt obligations + is able to invest that debt in high return ventures like AWS, Amazon logistics and core commerce.
From a pure balance sheet and valuation perspective, I prefer Alibaba at the moment by the way. It's got a better balance sheet (cash to debt ratio), but this is normal since Alibaba is far more asset light than Amazon.
Another business you can look at with similar debt loads is Costco. The balance sheet has 28b in cash and about 24b in current debt. But Costco runs a capex heavy business. It has to spend a lot of money upfront building out new warehouses which then mature over a 5-7 year period. They are able to fund this from compounding cash flow in each store.
Basically my point is : Debt is necessary and good as long as you can fund it easily with growing cash flow + your own Free Cash Flow continues growing after paying the monthly debt repayments. This means your ROIC is higher than debt obligations.
You make very salient points. I’m clearly in the minority here. What I’m saying is that what they describe as FCF isn’t actually FCF. It’s money they either owe merchants (this isn’t their cash, it’s a very temporary float plus subsidized shipping cost) or, with AWS, selling more and more server space on servers that are never paid for (AWS is also becoming a commodity fast!).
i like the 100 to 10 x 10 picture... this is decentralization. the benefits can be just as good if not better. compounding makes this possible.