During last ~10 years, nearly all high-end mountain bikes became 1x. Front derailleur seems to be gone. In road bike world this has not happened.
Why so? MTB gear range is typically larger than road one. In 3x era, MTBs could have 22-32-44 chainrings and 11-34 cassette, giving range >6, while road bikes might come with 53-39 front and 12-27 rear, ranging around 4. I'd expect that lower gear range allows for an easier 1x setup. What reasons could be there to not become 1x?
I’m going to state things a bit differently from the other answers.
On performance road bikes, you want to have small steps between at least the major gears and a wide range. Going 1x with even 12 cogs in the rear means some combination of less range or bigger jumps.
MTB suspension design improved tremendously when designers didn’t have to worry about the existence of the front derailleur. Road bikes don’t benefit from suspension systems. Their tires provide enough of it. Road bikes don’t gain much of any benefit from losing the front derailleur—there is a small aerodynamic gain, but part of it may be offset by higher drivetrain friction in the smaller cogs, and in the bigger cogs due to a poorer chainline (as compared to being in small ring big cogs on a 2x when climbing). It’s also considered best practice to install a chain guide to make sure the chain doesn’t fall off, which further offsets the aero gains.
However, there are niches where 1x will catch on. For dedicated cyclocross bikes, a lot of people will consider 1x. Time trial bikes can clearly benefit unless the course is hilly enough. Criterium racers with a dedicated race bike might also consider 1x. Some specific road courses can call for a 1x, or some people may live in really flat areas. But these are all niches.
Relatedly, if Classified’s geared hub takes off, more people would probably adopt it and run a single ring up front. However, the cost is high and the alternative (run a front derailleur) is quite effective. Therefore, I'm not sure how much take up Classified will achieve.
For historical reference, Aqua Blue Sport raced 1x11 SRAM drivetrains in 2017 and 2018. Their bikes were designed only for 1x. The team folded. The causes were multifactorial, but it seems like the big gaps at the top of the cassette may have contributed by leaving riders more fatigued. Some of the issues are discussed in this blog post. More recently, some teams have started to experiment with 1x in only certain races. Lizzie Deignan won the 2021 Paris Roubaix Femmes on a 1x12 drivetrain. Primož Roglič won the uphill time trial of the 2023 Giro d'Italia on a 1x12 as well.
Last, there are few 1x performance road bikes. However, a 1x drivetrain is much simpler for beginners. Experienced cyclists may scoff, but consider this: you and all your colleagues adapted to 2x, but you are not seeing the people who got turned off by 2x. It may not be the single deciding factor in a beginner leaving the sport because many failures have multifactorial causes, as above. However, it is an added stress point. I suspect many entry-level drop bar bikes will come in 1x in the future. Shimano's new CUES lineup of groups targeted at lower-end prices will be offered in 1x or 2x with 9-11 speeds. SRAM's Apex AXS and mechanical groups, which are equivalent-ish to Shimano Tiagra, are only offered in 1x.
It starts to appear on the high-end range, and is common for some very specific applications (time trial for example). A commonly heard critic of 1x transmissions is that there are not enough steps to be in an optimal ratio (for performance reasons) which is more important on road bikes than off-road bikes. Being in the optimal ratio is also more doable in practice on road, as there are less sudden changes of gears required. While off-road, conditions (slope, roughness) can change fast, and you are in practice almost never in the optimal ratio, and the ability to change fast is more important, so 1x drivetrains are more relevant in this application.
Another draw back of 1x transmission is to require the use of 10T (or 9T) small sprockets to have enough range, that are less efficient than the 11T ones.
That being said, 1x road transmission start to be used by some pro racers, on flat stages where the range offered by 2x is not useful - especially on stages like Paris-Roubaix where cobblestones require better chain retention, where 1x have an advantage.
Also, one of the reason of the demise of 2x in MTBs is linked to the increase in share of full-suspension MTBs (even in cross-country, that was traditionally dominated by hardtails): space is really at a premium around the bottom bracket of full suspension bikes, so removing the front derailleur simplifies the design.
EDIT: added some insights from this article (use by the pros, efficiency)
Unlike some innovations, electronic shifting seemed to take off faster on road bikes. That can be set up to allow the simplicity of sequential shifting that's inherent to 1x,but with more ratios.
MTBs also usually want lower ratios, and can get away with their only chainring being small. Their 1x setups lose out a lot at the top end. My old-school 29er hardtail has a highest gear of 46/11 or about 4.4:1. That works well for me, riding roads and gravel to reach trails or bridleways, but isn't going to be popular with those who only ride where an MTB is absolutely necessary. A typical 1x chainring of 32T gives less than 3:1 with an 11T sprocket, only just over 3:1 with 10T that wears really fast. It's even worse if that's also on 27.5" wheels. A road bike might want 50/11 at the top. To get the low ratios a typical rider might want in hilly country, you'd then need a 50T sprocket (only a little smaller if your highest gear is say 46/10). With several massive sprockets to give a range of low gears, you're not even saving weight.
Finally on MTBs 1x avoids some related failure modes, such as the front derailleur filling with mud. Done well, 1x should be less prone to a dropped chain (it's not always done well, and then dropping the chain can be more likely). More, perhaps larger, chainrings are also prone to getting bashed (though I've never worked out quite how I bent the smallest of my 3 without the others getting damaged, on a slippery local blue trail).
The motivation for 1x came from frame designers, not riders. MTB soft tail frames are extremely crowded around the bottom bracket. Suspension pivots need to strong and close to the crank. Removing the derailleur, they freed up space for better designs and lighter, stronger and stiffer bikes. 1x was also cheaper to manufacture and install.
Riders were sold on the benefits of 1x by marketing. Keep in mind early 1 was something like a 11-40tooth - nowhere near enough range for typical mountain bikers. The pros were fit enough and trained enough to mush 40-34 up steep hills and spin 130rpm on 11-34 on the flat to make it work, but mortal riders were forced into a massive compromise by frame designers. The hype around early 1x has to be one of the most significant marketing coups in product design. Now with 10-50 cassettes, the range problem is solved and 1x suits most riders better than 2x or 3x (at least until the cassette wears out and needs replacing).
Road bikes don't have the constraint around the BB and therefore no frame design benefit going to 1x. Road 1x has not been pushed onto roaders like MTB 1x was. Other have covered the benefits and drawbacks of why roadies are less likely to benefit from 1x than MTB. Even today, there is little real benefit in 1x for a road rider. If you consider cost of replacement cassettes (considered a consumable part), 2x or 3x has a significant benefit for un-sponsored riders.
It's really a marketing question. Salesmen haven't figured out a way to hype 1x. You do find 1x on work road bikes—city bikes and folding bikes (or time trial bikes for that matter)—just not on play bikes. People who bleed their brake cables aren't going to embrace the simplicity and easy maintenance of 1x. That's for people who use their bikes on the job. And it's hard to package it to sell: a new rear derailleur with a deep pitch but less wrap? a wide range cluster with subtly different spacings? a crank with a subtly optimized chain-line? Anyone with a handful of spacers and some Allen keys can try 1x for themselves.
A personal anecdote that will date me: Back in the 10-speed era, I crossed the continent. I set up 1/2 step gearing that gave me about 38" to 100" gears because I could use a prestigious Nuovo Record crank, and believed cross over gearing with common Japanese "touring doubles" (now successfully re-marketed as "compacts") was inherently wasteful. My low was higher than current racing lows, but we didn't spin uphill and state highways are graded for trucking. By the end of the trip, I wasn't bothering with double shifting on the lows and essentially using 7 gears. In retrospect, cross over gearing would have been simpler and just as effective.
My point is not that you need few gears. I love smooth continuous cadence enhancing gearing. Rather that one should gear to purpose.
Gear and cadence define speed, but external conditions—headwind, tailwind, uphill, downhill, scorching heat or bitter cold—define how much wattage you can output and how much you need. Since that is not predictable, you want gearing to offer you a small, consistent increase or decrease in effort wherever you find yourself on the range. However, one tooth steps at the high end won't match one tooth steps at the low. This is partly because of the math: 12 to 13 is the same jump as 24 to 26; And party due to going fast in high gears and slow in low gears. Usually this means you are fighting the wind in high gear and gravity in low gear. Effort against gravity is constant. Against the wind it is the square of the velocity. Small changes in speed in the wind, will require twice the change in effort. In other words, a 10% increase in gearing at the high end will be equivalent to a 20% increase at the low end. Of course, there is always a mix of wind, road, and gravity resistance so the difference will be somewhat less in practice. This all works out to 2, 3, and 4 tooth gaps in the bigger cogs of 11-30something clusters, which is why they are sold that way.
The limiting factor in 2x crossover is the front derailleur. Empirically a 14-16 tooth jump is the reasonable max from both a mechanical and physiological view. That means your gearing will cross over after about five gears. Thus, ten tightly spaced gears at the top, the remainder stretched out to cover the lower part of the range. Each extra gear on the cluster will add to the overlap with the big ring, and to the density of the low gears with the small ring. A little overlap is useful as it allows you to delay the double shift till more convenient. With nine speeds you make some choices. Thirteen is an embarrassment of riches.
With a 13 speed cluster, 1x is quite reasonable. Sure, it doesn't give you the 18 speeds that a 2x cross-over would, but you can start stretching out the gears sooner, and you will miss less that it first appears—while saving a bit of weight and that inevitable double shift.
I think the industry hit a high point when Campagnolo repackaged their cross-over triple with a 30 tooth low and rebranded it a "road triple". The trick is using the third ring to forgo cogs in the back. You can match the density and range of a 2x11 using an 8 speed cassette, while benefitting from greater durability, reliability, and tolerances of an 8 speed. The chain-line is better (not, as often stated, worse), the weight penalty is less than it appears as the cluster is smaller, the chain wrap only a link larger, and mostly comes down to an extra ring and bolt set. And shifting is buttery if you sensibly use a bar-con and, ahem, a chain watcher.
All this assumes cross-over patterns which makes sense with dense cassettes and mechanical shifting. Who needs to memorize double shift patterns? But it's trivial with a chip and apparently the pros are already using Alpine (X and 1/2 step gearing) pattern sequences programmed in. Not sure what happens if you hit the shift button three times, but I'm surprised it isn't being pushed more. I think the future holds 1x, 3x, or electronic 2x.
Roadies are terribly traditional.
Example - Disk brakes are basically standard on MTBs, but there are still bikes in the 2023 TDF that are rim braked. Team Jayco Alula from Australia, are riding the Giant Trinity Rim brake bike in the TDF TT in 2023.
Rim brake is behind the fork, and on chainstay (hidden behind chainring)
Coincidentally, this appears to be a 1x setup, because I can see no derailleur. Presumably the TT is a flat race and the aero advantages are worth it.
So a 1x road bike will come, but for pro racing will need a 53:11 gear and enough gear choices to keep the rider comfortable. I imagine we might see them in completely flat-land races, but the riders might return to 2x for any stages/races with significant climbing.
Two words: efficiency and weight.
Road bikes are ridden fast. To go fast, you need high gearing. You can obtain high gearing by either a very small sprocket in the rear or a very big chainring in the front.
The problem of a very small sprocket in the rear is efficiency. While current cassettes have 11-tooth sprockets in case you want to pedal really fast downhill, in typical environments you are more likely to use a 15-tooth sprocket. 15-tooth sprocket is reasonably efficient. 11-13 tooth sprockets are not.
So because you want to limit your flatland gearing to at least 15+ tooth sprocket, it means the chainring in the front must be big enough to have regular pedaling cadence at typical road bike speeds. You usually don't see road bikes with smaller than 50-tooth chainrings, and many have 53-tooth chainrings.
So let's assume you have a 50-tooth chainring. Let's also assume that you want the ability to pedal up very steep hills, with approximately 1:1 gearing.
While today you can find 11-51 tooth cassettes allowing 1:1 gearing with 50-tooth chainring, that's far from optimal. Firstly, the chain needs to be sized to work on the 51-tooth sprocket and chains are made from steel (heavy if long). Secondly, the biggest sprockets, 51-tooth and the ones near it, are heavy as hell since sprockets are typically made of steel.
So a second option is to forget 1x and put a smaller, let's say 34-tooth, chainring and a front derailleur with 11-34 cassette. The chainring is most likely made of aluminum since chainring wear doesn't cause skipping like rear sprocket wear does. The single 34-tooth aluminum chainring is not as heavy as a whole bunch of big steel sprockets. Front derailleur does add some weight, but not as much as you would add by making the chain really long and having a whole bunch of big steel sprockets.
Also you save a bit of weight from the rear derailleur since it doesn't need a very long cage.
I won't say anything that wasn't mentioned in other answers, but I want to demonstrate how shifting works on road.
The importance of smooth close-gap shifting cannot be overstated. Significant and sudden changes in cadence are not only bad for performance, they are very annoying. Unlike MTB, on road, sudden changes of conditions that warrant a large instant shift are very rare (and if they happen, shifting the front often achieves exactly that). In contrast, starting a 1% climb changes the speed just enough (2-3 km/h) to need a shift.
To maintain the cadence in the optimum range, say, 90 to 100 RPM requires no more than 10% steps in shifts. That necessitates single-tooth gaps up to about 19. The ideal road cassette (for traditional 11-speed systems) is therefore 11-12-13-14-15-16-17-18-19-21-23. The last 1-3 sprockets can be upped a bit if extra range is needed: at the extremes, smoothness of cadence can be sacrificed. But it's still impossible to accommodate it for 1x. Not even close. Even 2x struggles to achieve enough range with such an ideal cassette, and compromises have to be made.
I have a hybrid with slightly wider-spaced cassette than my road bike, and often take it on the same commuter route. And it's really, really annoying not to have 16 and 18 - even though they make only 12-13% gap.
