By Benedict Evans from A16Z
We all know, I think, that there are now far more smartphones than PCs, and we all know that there are far more people online now than there used to be, and we also, I think, mostly know that big tech companies today are much bigger than the big tech companies of the past. It’s useful, though, to put some real numbers on that, and to get a sense of use how much the scale has changed, and what that means.
This change is even more striking if you shift the timeline. If you compare GAFA in their current dominance with Wintel in their period of dominance, you see not a 3x difference in scale but a 10x difference. Being a big tech company means something different now to in the past.
Scale means these companies can do a lot more. They can make smart speakers and watches and VR and glasses, they can commission their own microchips, and they can think about upending the $1.2tr car industry. They can pay more than many established players for content – in the past, tech companies always talked about buying premium TV shows but didn’t actually have the cash, but now it’s part of the marketing budget. Some of these things are a lot cheaper to do than in the past (smart speakers, for example, are just commodity smartphone components), but not all of them are, and the ability to do so many large experimental projects, as side-projects, without betting the company, is a consequence of this scale, and headcount.
On the other hand, that the market is big enough for four tech giants, not just one (Wintel) partnership, means we have four companies aggressively competing and cooperating with each other, and driving each other on, and each trying somehow to commoditise the others’ businesses. None of them quite pose a threat to the others’ core – Apple won’t do better search than Google and Amazon won’t do better operating systems than Apple. But the adjacencies and the new endpoints that they create do overlap, even if these companies get to them from different directions, and as consumers we all benefit. If I want a smart speaker, I can choose from two with huge, credible platforms behind them today, and probably four in six months, each making them for different reasons with different philosophies. No-one applied that kind of pressure to Microsoft.
How do the mice do when there are four elephants fighting it out? As we saw with first GoPro and now perhaps Sonos, if you’re riding the smartphone supply chain cornucopia but can’t construct a story further up the stack, around cloud, software, ecosystem or network effects, you’re just another commodity widget maker. And the aggressive competition in advertising products from Google, Facebook and now to some extent Amazon has taken a lot of the oxygen away from anyone else.
Looking beyond the scale and the network effects, though, there’s a difference in character. Google, Facebook and Amazon are still controlled by their founders, and they’re aggressive street fighters. All of these companies have the benefit of twenty years more history – they saw what happened to Microsoft, and Yahoo, and AOL, and MySpace. So, they will disrupt themselves, and they will act. The shift to mobile was a fundamental structural threat that unbundled Facebook – the founder spent over 10% of the company to buy the most successful unbundlers and, as importantly, didn’t smother them after he’d bought them, unlike most large acquirers of disruptive companies. Just as index funds don’t work if everyone’s an index fund, you could propose that ‘Disruption’ doesn’t work if everyone’s read the book, and everyone has. This, to repeat, is compounded by scale, both for strategic shifts (such as chips) and for people: the big tech companies have hired a huge proportion of the stock of academic machine learning researchers in the last few years, paying huge (cash!) salaries and offering both freedom and the change to deploy something real to billions of people. Not everyone takes the money (Evan Spiegel didn’t), but not every engineering or product genius wants to be an entrepreneur and sleep like a baby – to wake up every hour and scream.
That is, the new platforms have created unprecedented opportunities – 3bn smartphone users today are an unprecedented ‘white space’ for company creation. But they often expand into that space themselves, both on their own platform (an old problem, of course, seen at Microsoft in the past), and using their leverage to try to seize any new white space that opens up and may be a threat (Facebook buying WhatsApp and Instagram and trying to squash Snap). These impulses are not new: Microsoft was hardly a friendly competitor and Intel’s Andy Grove famously said that only the paranoid survive, but the ability to act on them is different – GAFA can do more and much bigger things than Microsoft ever could. My colleague Chris Dixon calls them ‘super-evolved organisms’ – they’re arguably more aware of where threats may come from, and certainly more able to respond.
On the other hand, both in tech and the broader economy, large, dominant companies don’t last. You lose the market or the market becomes irrelevant. Nokia had close to half of the mobile handset market a decade ago and lost it all; IBM still has the mainframe market but no-one cares. Few people can predict where the change will come from, but it does come. GAFA are very visibly conscious of that – Google experiments with everything, Apple is working on cars and mixed reality, and Facebook bought not just Instagram and WhatsApp but Oculus. But then, Microsoft was working on smartphones and mobile devices 20 years ago, and now it’s killed Windows Mobile, acknowledged that the PC is going the way of the mainframe and, like IBM, has to make its way in a market shaped by other companies. There probably won’t be a technology that has 10x greater scale than smartphones, as mobile was 10x bigger than PCs and PCs were bigger than mainframes, simply because 5bn people will have smartphones and that’s all the (adult) people. There will be something, though, and though ’something will change, but we don’t know what’ is an unfalsifiable point, so is ‘nothing will change’, and I know which side of that argument I find more likely.