Gross Margins Determine Strategy
Tech companies are moving from zero marginal cost revenue to high marginal cost revenue. The impact will be large. I will examine how high gross margins have affected the behavior of Big Tech, and then connect it to how startups will scale in the future.
In Big Tech, gross margin has wide dispersion, ranging from more than 80% for Meta to less than 50% for Apple and Amazon.
This gross margin disparity determines company strategy. As AI proliferates, all companies, especially startups, will report lower gross margins. And it will increase focus on operational excellence.
Those with higher gross margins tend to make more acquisitions and take more risky bets. This is personified with Meta and Microsoft. They go all out to protect their economic franchise.
Those with lower gross margins tend not to make splashy acquisitions. These companies have to operate very efficiently and focus less on acquisitions. You can clearly see this behavior in Apple and Amazon.
These extra points of gross margin are basically viewed as additional poker chips that can be used to protect the economic franchise.
Big Tech Case Studies
Meta
Meta's high gross margins imply a valuable economic franchise, with network effects as the core competitive advantage. With revenue of 160B a year, the 30-point difference in gross margin between Meta and Amazon means that Meta has 48B available to protect that high-margin income stream. And they’ve done that with Reality Labs and now with their AI initiative. Anything to protect their global walled garden.
Apple
Apple’s gross margin is the result of deep interaction with the physical world, which always carries the residue of cost of goods sold. However, this hardware ecosystem has turned into a platform that delivers high-margin revenue, and you can see that mix shift in Apple's gross margins over time. Still though, Apple has less money to deploy by virtue of its lower gross margin. What capital they do spend goes to share buybacks. Apple clearly believes more capital expenditure or acquisitions won’t do much to protect its moat.
Amazon
Amazon’s sub-50% gross margin comes from a different place. They must support a global procurement network to facilitate next-day shipping to billions of people. This platform has become a moat, but one that requires continual investment. Even AWS, their other major business, accepts lower gross margins. They seem to believe a dollar in reduced gross margin is a better investment than anything else.
I was surprised to see Google’s gross margin as low as it is, but then remembered they pay 20B to Apple every year. With 350B of annual revenue, that is a direct transfer of 6 points of revenue to Apple. Google’s scale is notable—their services cover a wide surface area and penetrate deep into every internet crevasse.
Microsoft
Microsoft, with around 70% gross margins, has a more complicated story. The financial picture is more varied, and I can’t break down the source of their supranormal margins in one sentence. But it’s clear they are highly acquisitive in the interest of protecting their economic franchise.
Gross Margin vs Operating Margin
Because of the aggressive spending of the Big Tech companies with higher gross margins, operating margin becomes less dispersed. The total level of spend as a percentage of revenue converges.
The clear outlier is Amazon, who has consistently accepted low gross margin and low operating margin as a key competitive strategy.
Here's the full picture:
| Company | Revenue | Cost of Revenue | OPEX | Profit |
|-----------|---------|------------------|------|--------|
| Meta | 100 | 20 | 40 | 40 |
| Microsoft | 100 | 30 | 25 | 45 |
| Google | 100 | 40 | 30 | 30 |
| Apple | 100 | 50 | 20 | 30 |
| Amazon | 100 | 50 | 40 | 10 |Startups in the AI Era
All AI startups are going to resemble Amazon’s AWS or Apple’s hardware-software stack more than Meta’s ad-driven engine or Microsoft’s enterprise SaaS. Unlike traditional software, AI has a tangible marginal cost.
Every token generated is powered by electrons, and those electrons are still produced from carbon. Its economics are grounded in atoms: chips, cooling, energy, and latency.
In that sense, AI is transmuted electricity, and electricity is transmuted carbon.
This energy intensity structurally caps gross margins and forces AI startups to operate more like logistics or hardware companies.
Historical Startup Economics
Startups historically burned money via the OPEX line. Pre-AI, the income statement of a scale-up tech company looked like this:
| Year | 1 | 2 | 3 | 4 | 5 |
|------------|-----|-----|-----|-----|------|
| Revenue | 1 | 2 | 4 | 8 | 16 |
| CoR | 0.1 | 0.2 | 0.4 | 0.8 | 1.6 |
| OPEX | 4 | 5 | 6 | 7 | 8 |
| Profit | -3.1| -3.2| -2.4| 0.2 | 6.4 |
| Cum Profit | -3.1| -6.3| -8.7| -8.5| -2.1 |Now, it looks like this (same OPEX, but higher cost of revenue):
| Year | 1 | 2 | 3 | 4 | 5 |
|------------|------|------|------|------|-------|
| Revenue | 1 | 2 | 4 | 8 | 16 |
| CoR | 0.75 | 1.5 | 3 | 6 | 12 |
| OPEX | 4 | 5 | 6 | 7 | 8 |
| Profit | -3.75| -4.5 | -5 | -5 | -4 |
| Cum Profit | -3.75| -8.25| -13.25| -18.25| -22.25 |The Game Has Changed
The rules of the game have changed. You no longer have to outgrow your fixed cost base. There is no fixed cost base. It's variable now.
As I type, every Big Tech firm, except Apple and Amazon, is caught up in the AI CapEx game in a race to protect their economic franchises.
And every VC is shoveling money into startups designed to pierce the armor.
As always, the biggest investment gains will come from businesses that can deliver services at maximum effectiveness (high gross margin) and exhibit efficiency as they grow (operating leverage).