要靠5G重回移动市场，英特尔都在做什么

This implies that Qualcomm's upcoming Gobi modems are nowhere near the speeds of 5G. Investors can anticipate uplink and downlink speeds that are substantially better than the projected performance of Snapdragon X12 LTE (Gobi 9x40/9x45 series) upon the introduction of 5G speeds. This moves my timeline for when I expect 5G and the speed that I anticipate from 5G. Currently, the most advanced mobile modem has peak speeds of 450 Mbps on the downlink and 100 Mbps on the uplink. If we were to apply a conservative estimate on how 5G would transform uplink and downlink speeds, we can anticipate 45 Gbps assuming 100x performance scaling. At 1,000x performance scaling we can anticipate 450 Gbps of uplink speed, which will likely be the speed of later generation 5G modems. To put 45Gbps in context, we can compare it to the distributed denial of service attack that hit a large Chinese e-commerce company back in 2011. The server had to continually deny requests from 250,000 infected computers. So by 2020 wireless speeds on a mobile handset will be equivalent to the bandwidth usage of a large-scale network attack from 2011.

Talk about science fiction becoming a reality.

This has interesting implications for network infrastructure because there are two ways to essentially increase speeds. One way is to improve the available spectrum which allows the mobile Internet to travel along more frequency bands. However, the other way is to increase the speed at which data moves within the available spectrum.

Another way of thinking about it - spectrum is the available lanes on the Internet freeway, whereas an improvement to uplink and downlink refers to the speed at which data travels along the Internet freeway. So, to accommodate for more devices accessing the Internet, the implementation of 5G will have to take place, as there's very little spare capacity on 4G networks currently. Verizon (NYSE:VZ) conveys that they can convert more of the available spectrum for 4G, but it's really a short-term solution for accommodating the long-term data needs of 50 billion projected devices. Of course telecoms have to build a network to accommodate for the projected bandwidth needs of more device, which is why I'm very hesitant to recommend the lower-tier networks because their capital spending doesn't really match the long-term data needs of mobile users. Much of this increase in activity will exist at the edge of the network, which will put pressure on content delivery networks to sustain last-mile connections, which currently accounts for a lot of the activity that currently takes place on the web.

Of course, Intel isn't out on a limb making up these numbers, as Verizon mentioned on its second quarter earnings call:

There's a lot of things will bring efficiency to the network. There is C-RAN out there. We are in the initial - obviously 5G is being talked about in the industry. Of course, Asia is involved in 5G and of course we will start to get involved in the standard setting around 5G, so there's a lot happening in this industry for technology standpoint.

If the mobile carriers are also getting involved with the transition to 5G it's safe to bet that the technology is about 5 to 10 years out, as it's extremely unlikely that 4G will sustain the projected explosion in data usage and the current pricing model isn't very appealing to consumers and offers limited viability for growth going forward. Therefore, while the transition to 5G is capex intensive, i