Why it matters: After looking at revenue and operating income per employee for big semiconductor companies, we thought that was a fun exercise, so we have looked at another dozen tech companies in various sectors. Broadcom and Apple are in a league of their own – it is good to have a software or licensing business.
During Qualcomm's earnings call earlier this month, one analyst pointed out that the company has twice as many employees as its peers. That thought wormed its way through our brain, and so we put it to the test.
Editor's Note:
Guest author Jonathan Goldberg is the founder of D2D Advisory, a multi-functional consulting firm. Jonathan has developed growth strategies and alliances for companies in the mobile, networking, gaming, and software industries.
We looked up the employee count, revenue and operating income of the five largest semiconductor companies for their most recent fiscal year. Since we had so much fun with that, we extended the search to include the major Internet companies. We also had a run at telecom equipment makers recently, and so we added them into the mix.
And since some of those companies have large warehouse and retail footprints, we added a couple of major retailers, too. We then calculated revenue and operating income per employee of the whole set.
| Company | Employees | Revenue ($ b) | Op Profit ($ b) | Rev/Employee ($) | Op/Employee ($) |
|---|---|---|---|---|---|
| Amazon | 1,684,853 | 513.98 | 12.25 | 305,041 | 7,259 |
| AMD | 25,000 | 23.60 | 1.26 | 944,040 | 50,560 |
| Apple | 164,000 | 539.30 | 108.95 | 3,288,659 | 664,725 |
| Broadcom | 20,000 | 26.28 | 14.23 | 1,313,850 | 711,250 |
| Cisco | 83,300 | 57.69 | 10.08 | 692,559 | 121,224 |
| Ericsson | 105,000 | 23.612 | 2.35 | 224,876 | 22,381 |
| 75,964 | 116.61 | 28.94 | 1,535,225 | 381,225 | |
| 190,000 | 282.84 | 74.84 | 1,488,611 | 393,905 | |
| Huawei | 114,000 | 92.38 | 6.07 | 810,342 | 53,295 |
| Intel | 131,900 | 63.05 | 2.33 | 478,014 | 17,665 |
| Microsoft | 221,000 | 211.92 | 88.52 | 958,891 | 400,557 |
| Nokia | 86,896 | 26.50 | 2.47 | 304,974 | 28,379 |
| Nvidia | 26,196 | 26.67 | 4.22 | 1,018,247 | 161,246 |
| Qualcomm | 50,000 | 35.82 | 7.79 | 716,400 | 155,760 |
| Target | 400,000 | 107.59 | 3.85 | 268,970 | 9,625 |
| TSMC | 73,090 | 70.48 | 3.78 | 964,322 | 51,659 |
| Wal Mart | 2,300,000 | 605.88 | 20.43 | 263,427 | 8,882 |
Already this is fairly interesting, but we will break it down by sector to make it more readable. The tables below show subsections of the data above. We have grouped them by sector and then ordered them first by revenue per employee...
Semiconductors
| Company | Employees | Revenue ($ b) | Op Profit ($ b) | Rev/Employee ($) | Op/Employee ($) |
|---|---|---|---|---|---|
| Broadcom | 20,000 | 26.28 | 14.23 | 1,313,850 | 711,250 |
| Nvidia | 26,196 | 26.67 | 4.22 | 1,018,247 | 161,246 |
| TSMC | 73,090 | 70.48 | 3.78 | 964,322 | 51,659 |
| AMD | 25,000 | 23.60 | 1.26 | 944,040 | 50,560 |
| Qualcomm | 50,000 | 35.82 | 7.79 | 716,400 | 155,760 |
| Intel | 131,900 | 63.05 | 2.33 | 478,014 | 17,665 |
We should not have been surprised to see Broadcom lead both of these tables for semis. In part, their data is distorted by the fact they sell a lot of software which tends to boast larger productivity figures. Nonetheless, their figures stand out. For starters, they have 20,000 employees (possibly less after a recent headcount reduction there). They are also optimized for profitability, befitting a private equity portfolio. Nvidia also stands out on this list. And we are looking at FY22 revenue, so their recent massive earnings upside are not fully reflected here. For the Nvidia bulls out there, imagine what the company can do if it starts to boost its software revenue and approach Broadcom's profitability levels.
To answer the question which kicked all of this off, Qualcomm does indeed have 50,000-ish employees, double those of Nvidia and AMD, but on a profitability scale they score rather well, coming in third. The company has to spend a lot on R&D but that effort is rewarded via their highly profitable licensing business. That being said, Qualcomm seems very generously staffed. As a counterfactual, imagine what Broadcom's results would look like if they had succeeded in acquiring Qualcomm. And Qualcomm still has the jets…
Intel comes in at the bottom of both rankings. In fairness, their results are depressed lately because of other problems. If (when?) they can really turn the corner and regain some of their profitability they will likely climb. In addition, their figures are also burdened by their manufacturing workforce which is fairly sizable. On the other hand, TSMC has 60,000 fewer employees, which likely means that some part of Intel is still overstaffed.
FAAMG
| FAAMG | Employees | Revenue ($ b) | Op Profit ($ b) | Rev/Employee ($) | Op/Employee ($) |
|---|---|---|---|---|---|
| Apple | 164,000 | 297.39 | 108.95 | 1,813,366 | 664,323 |
| 75,964 | 116.61 | 28.94 | 1,535,056 | 381,023 | |
| 190,000 | 282.84 | 74.84 | 1,488,611 | 393,905 | |
| Microsoft | 221,000 | 211.92 | 88.52 | 958,891 | 400,557 |
| Amazon | 1,684,853 | 513.98 | 12.25 | 305,061 | 7,269 |
The one surprise for us on this list is Google. The last time we ran this analysis (~10 years ago) they were by far and away the leader in this comparison.
Turning to the hyperscalers, it should come as no surprise that Apple sits at the top of both rankings. We should point out that they are slightly less productive than Broadcom when it comes to profit per employee. Regardless, Apple is clearly a different kind of company. By the same logic, it is not surprising that Amazon sits at the bottom as it has a massive logistics workforce and a historic aversion to being too profitable. The one surprise for us on this list is Google. The last time we ran this analysis (~10 years ago) they were by far and away the leader in this comparison. They have added a lot of employees since then and are yet to see the benefit of that.
Networking
| Company | Employees | Revenue ($ b) | Op Profit ($ b) | Rev/Employee ($) | Op/Employee ($) |
|---|---|---|---|---|---|
| Huawei | 114,000 | 92.38 | 6.07 | 810,342 | 53,254 |
| Cisco | 83,300 | 57.00 | 15.03 | 684,250 | 180,444 |
| Nokia | 86,896 | 26.50 | 2.47 | 304,974 | 28,379 |
| Ericsson | 105,000 | 23.612 | 2.35 | 224,876 | 22,381 |
Turning to communications equipment, the surprise leader in revenue per employee is Huawei. These are 2022 numbers, so prior to their recent 'turnaround' and at the nadir of the US restrictions' impact.
On the other hand, they are much less profitable per employee than Cisco, another company benefiting from a shift to software and licensing. And in fairness, Huawei is not publicly traded so their publicly-reported financials may not be based on the same accounting basis as the other companies. Nokia and Ericsson round out the bottom, again not a surprise.
Retail
| Retail | Employees | Revenue ($ b) | Op Profit ($ b) | Rev/Employee ($) | Op/Employee ($) |
|---|---|---|---|---|---|
| Apple | 164,000 | 297.39 | 108.95 | 1,813,366 | 664,323 |
| Amazon | 1,684,853 | 513.98 | 12.25 | 305,061 | 7,269 |
| Target | 400,000 | 107.59 | 3.85 | 268,970 | 9,620 |
| Wal Mart | 2,300,000 | 605.88 | 20.43 | 263,427 | 8,882 |
Finally, recognizing that Apple has a large retail footprint and Amazon has its logistics network, we wanted to see how the two companies compare to retail giants Walmart and Target. Again, Apple is off the charts, in hindsight it was probably unfair to put them on the list. Intriguingly, Amazon is the least profitable per employee of the group.
... software is still eating the world.
So to answer the question which set all of this off, Qualcomm does have a larger workforce than its semiconductors peers, but makes up for some of that through its licensing business.
We also came away with two big takeaways: first, business cycle matters a lot in these comparisons. Intel and Google's decline over the years is telling, and those forces matter more than headcount. Second, software is still eating the world. Having a high-margin software or licensing business is a great model if you can find one that works.