Can we restore the glory days of tech, when computing and networking spending were both growing? There are demand-side opportunities in play, as I said in yesterday’s blog. Tomorrow could bring new stuff. Today, the supply side of networking and the rest of IT is hamstrung by commoditization. There has been no significant differentiation, and no significant new business cases to drive new spending. As a result, buyers have demanded more for less, and in many cases “more” means greater “current power” to do the same job, which ends up meaning spending less too.
To complicate matters, commoditization is combined with the fact that seller companies are focused primarily on sustaining their share prices, and for that Wall Street looks primarily to the current quarter and not the far future. Sellers then are not inclined to push long-term changes that impact short-term sales. They’re cutting staff, cutting R&D, because they’ve not been rewarded much for innovation. A lot of challenges, but not all the supply-side providers are equally impacted. There are differences in the inertial barriers to responding to new opportunities among suppliers, enough to divide them into three sectors.
The fastest and best of these is application and middleware software. Software can be produced by a couple programmers in a basement. A software startup doesn’t require a boatload of money. Software investment by buyers can typically be controlled, so there’s not a major barrier to adopting stuff from an unfamiliar source. Software is what turns hardware into something other than a space heater, it’s the functionality wellspring of tech. All that means that software could be expected to respond to opportunity changes, reflect a new demand paradigm. All that software lacks is publicity (and there are plenty of skilled publicists) and some seed money (and VCs are despairing of any new opportunity).
The next sector is the hardware and platform software group. This is long-lived as a buyer asset, so there’s resistance to change. Manufacturing it is an undertaking, which means it requires a bigger organization with more capital. It can’t do anything without the application/middleware stuff, so not only is it a natural second-mover, it presents the insecurity of dependency in its own planning. What if the thing above doesn’t do its job? My whole strategy is crap. I’m ruined. But this sector is already feeling the pressure of commoditization, seeing stagnation in buyer opportunity. You can’t create Fortune 500 companies with a wave of the hand, and this sector wants big buyers.
Finally, there’s networking. Talk about being the bottom of the pile. If you can’t generate new experiences, you have nothing new to deliver. No new users, nothing new to connect users too, nothing people want to pay incrementally for. No TAM, no ARPU. Add to this a combination of higher-than-average depreciation cycles, and a culture that dates back to the era tall stools and workers whose sleeves are held up with elastic bands, wearing green eye shades, and you have networking. Inertial conservatism is a fatal disease. Which is why we have to assume something dire could happen here.
There is a way out for network vendors and operators, which would be to take steps necessary to accelerate the marriage of the new demand paradigm and the supply side of tech. They’re almost certain to avoid that notion like the proverbial plague. However, just as forces divided our supply side into groups, they now act to divide the network players, and this time the separation is based on the simple metric of market share.
Big players, the major incumbents, have a lot to lose when faced with change, and less to gain. Get smaller and the balance shifts to the opposite. Smaller players, particularly those who are either startups or who have created a successful but narrow niche for themselves, would see the same opportunity picture that the more agile software types see. They could even try to create a partnership, but even if all they did was watch for the software sector to move, it would be enough. They would almost surely take steps to exploit the new opportunity. Cisco won in enterprise networking because the big incumbent of the time, IBM, didn’t see where things were heading. The next Cisco will be created by the same thing, and by the way, I’ll have a bit to say about Cisco’s NVIDIA deal below, and even more in tomorrow’s blog.
For service providers things are a bit more complicated. They’ll be even more adverse to change than network vendors, their financial inertia is longer, and their capital demands are larger…at least in the traditional model. However, services are subdividing into “Internet” and “access to the Internet”, with the latter being divided into wireless and wireline.
There are plenty of ways for wireline access providers to get connected with the Internet, and the big question for the provider space is whether every little community will be served by one of the current giants or whether some will be too small to be interesting. In the EU and many Asian countries, densities are high enough that this isn’t a big issues, but in countries like Australia, Canada, and the US, we could easily see the big players hanging back on full empowerment of smaller communities. Local governments and smaller providers could step in, and this could create a kind of community-virtual-giant player. And since big players may resist shifting their larger customers from VPN to Internet and SD-WAN, the smaller ones might pick up that too. All this means that unless telco innovation ramps up very, very, quickly, the wireline part of the telcos are going to be nothing but utilities, driven to manage costs to be cheap.
What about wireless service? Spectrum issues will keep wireless from Balkanizing, and because of that the major telcos are likely to focus more and more on wireless, to the point where we are already seeing operators like T-Mobile that are nothing else, and we may even see wireline telcos deemphasizing what used to be their only access technology decades ago.
The problem with wireless is that you can’t grow humans by deploying antennas. For decades, wireless total addressable markets were growing globally. Today, in major economies, there’s very little TAM growth possible any longer. When IoT came along, telcos fell all over themselves trying to promote not the application technology facilitation that would have realized the true IoT opportunity we noted in the last blog, but the notion that an Internet of Things meant you could sell cellular service to Things in addition to people. Instant TAM growth. It was nonsense of course, but it illustrates how far telcos and their vendors will go to stick their head in the legacy-service-business sand.
This raises what I believe to be the big supply-side question for networking, and even for tech overall. Will we, will the market, actually do something about the current challenges? We actually have both demand and supply trends that would argue that progress in harnessing the smartphone as a means of projecting a very broad and very new class of experiences would be good for all. The problem is that broad, new, things are scary things, disruptive things. So is a slow slide into Chapter 11 or being acquired. We are now seeing, we are in fact mired in, the inertia of the current-quarter fixation. I am surviving, so I should accept that as the only positive needed. That I am slowly sinking is unfortunate, but trying to slow the sink rate might actually accelerate it. What happens if I tell my buyers that the next great thing is “X”, but that its value can’t be fully realized for half a decade? Might they sit on their wallets till that value is on the current table? The answer to these questions could come from four distinct places, but before I list them I want to briefly mention a big player who’s not on my list, Cisco.
Cisco just announced a deal with NVIDIA, and they’re touting it as a major AI initiative. In a way it is, but it’s really a marketing initiative and not a technical one. Cisco isn’t breaking new ground here, they’re on the defensive. I’ll blog in detail about that tomorrow. For today, I want to list out the players to really watch.
The first place is Apple. This is a company who owns a big chunk of the loyal, leading-edge-tech, consumers. They are placing a huge bet on AR/VR, the kind of bet that if they lose will surely sink their stock as much as anything has sunk it in the past, or even more. They are highly innovative and their loyal base gives them the opportunity to push their stuff into the market with less hard justification than other vendors might need. For them to fully realize AR/VR potential, they need a lot more than glasses, and they may be willing to do a lot of their own R&D and partner favorably with other innovators to get it.
The second is Meta. They have admitted that their own business model, based on extensive social network participation, is getting thin. The metaverse was their answer, but so far they’ve failed to get real traction with it. They likely believed that AR/VR glasses were a sufficient condition for metaverse success, and I think you could argue that they weren’t even necessary conditions. That honor goes to digital twinning. If Meta leans into that topic, they could move markets.
Then we have IBM. This is a company that’s had AI for a long time, but couldn’t seem to find the right angle for it. Now, they seem to have done that. IBM’s watsonx strategy has real potential, and if it could drive enterprise adoption of AI in multiple analytics business missions, it could bring AI to thousands of companies. Through Red Hat, IBM has the breadth to hit so much of the potential AI market that it could be dazzling, but they could have done that before, too. When somebody can and does not, you wonder whether there’s a cultural or other barrier in place.
And last? A place you may not have expected. HPE/Juniper. HPE could justify the acquisition in only two ways. First, they could believe that the economies that could be gained by consolidating the two workforces would be so enormous that the cost reduction would render both HPE’s computer business and Juniper’s network business safely profitable even in the face of commoditization. Second, they could see some symbiosis in play, an opportunity to present a single new driver to IT spending that would lift both the computer and network boats. If they think that the first justification will work, I think they’ll doom themselves. If they think the second will work, then they could advance the real-world integration of IT, networking, and the user that I think yesterday’s demand-side blog illustrates.
Four options, with four big “ifs” attached. Will any play out? Will they all? That’s the question that will determine whether tech booms again. Business as usual will otherwise yield results that have become all too usual. The most interesting outcome, of course, is if all three of these companies make a move, and the one thing the Cisco/NVIDIA did was raise the odds that might happen. It’s not causal, but an indicator.
Cisco is a fast follower, and so the fact that they did this deal, and did it when IBM’s stock was on the rise due to AI, and shortly after the HPE/Juniper deal and the Juniper AI-Native announcement, suggests that they believe that there’s a real chance enterprise AI will happen. That’s important because enterprise AI, unlike all the hype about generative AI and public models, is small enough to be widely applied. It’s safe to use on company data, even trade secrets. It’s small enough to deploy in many different missions. It’s probably a lot cheaper than public-model generative AI, so it doesn’t require much of a business case. It could make AR/VR and the metaverse into something truly transformational, if both Apple and Meta decided to take advantage of it, and if IBM or HPE/Juniper proved that it could be deployed in a practical way. This is the best outcome from a market perspective, and so it’s the one I hope for. We’ll just have to see what happens.
