I don’t mean to be dismissive of the question, but it’s mainly because transmitting is harder than receiving. To begin with, the RTL-SDR is based on an integrated circuit that’s intended for watching HDTV over the air. The market for broadband HDTV receivers is enormously bigger than the market for amateur transmitters, and justified the cost of developing receivers. Custom or Application Specific Integrated Circuit (ASIC) development generally runs into hundreds of thousands of dollars, perhaps half a million (and that’s if it’s right the first time). Plus, the chip designers thought of what the future features might be wanted in the HDTV receivers in the future and added digital data streams; called I/Q data (In-phase and Quadrature-phase) signals, which is what allowed the software developers (hackers) to use the chip as a broadband receiver. On top of all that, to be brutally honest, the receivers just aren't that good (toward the bottom of this post).
Then there’s the nature of transmitting itself. I know I’ve talked about this before, but transmitters (like all circuits) have an output impedance and antennas have an input impedance. A circuit theorem called “Optimum Power Transfer Theorem” says that for that to happen, they must be the same impedance. A convenient way to approximate the output impedance of a transmitter is the voltage squared over 2 times the power output ((Vcc^2)/(2*Po)), which then gets matched to the output of the amplifier circuit, which is generally 50 ohms. The higher the transmitter power, the lower the output impedance, which makes matching it harder, which means more parts, which leads to less bandwidth. The transmitter connects by transmission line to the antenna, often specified to 50 ohms as well. The antenna, in turn, can be thought of as an impedance transformation from the 50 ohms at its input to the 377 ohms impedance of free space.
The broader the bandwidth of the antenna (i.e., the frequency range over which the antenna presents a 50 ohm load), and the broader the transmitter output impedance match, the more broadband a transmitter can be.
If you’ve played with an RTL-SDR for any length of time, you have probably noticed that it seems more sensitive on some bands than others. That’s because the chances are you’re using a non-resonant antenna, like a rubber duckie or a short wire. These are intended to match around their electrical length (remember for a quarter wave monopole, 234/f gives the antenna length in feet when f is MHz) and the match gets worse as you tune further away. The RTL-SDR folks sell an upgraded version of the receiver with a collection of antennas, but they’re best from around the FM broadcast band up to around 1.5 GHz, although you’ll have to swap antennas and change their lengths while trying to listen to something.
You can do better. There are antenna designs that are much broader bandwidth than simple monopole “rabbit ears” antennas like the ones that RTL-SDR sells, but those designs cost more than the “rabbit ears.” The two that come to mind first are one called a discone; these tend to be operated in vertical polarization, and a log-periodic dipole array (LPDA), which has the advantage over the discone of offering antenna gain (directionality), and can be mounted in either vertical or horizontal polarization.
The antenna, then is less of an issue. A log periodic with a bandwidth of 10:1 (say 50 to 500 MHz) isn’t exactly bleeding edge technology and a 144 to 1300 MHz (9:1 frequency) transmitting Discone is under $50 from Amazon. If we can put a transmitter on the antenna that’s a broadband output, is that our transmit equivalent of the RTL-SDR? Nope. The SDR linked to above will receive 500 kHz to 1500 MHz, and the broadband antenna is over two orders of magnitude away from 500 kHz (288 times 500 kHz).
Do broadband 50 ohm output transmitter systems exist? Yes, but with two problems. First, you don’t want to pay for one that delivers power of as little as 25 Watts. These things go for thousands of dollars. How about a 1W (in FM) or 100mW SSB amplifier from 10 to 4200 MHz; that’s only $1500?
The second problem is transmitters have an inherent need for filters. While the receiver side will benefit from filters, the only one who suffers is you; the user. When transmitting, your signal can carry harmonics and spurious signals that interfere with other spectrum users. Interfering with other services can bring enforcement and big fines. In the US, we’re generally able to receive anything we can build hardware and software to receive, but this isn’t the case in other countries.
When I started thinking about this problem, I thought the enabling technology might be a DDS with a modulator on its output. Analog Devices has a line of products they call TXDACs (Transmit Digital to Analog Converters) that can generate the signal and produce a low power output (as in milliwatts). These signals need to be software generated but give two or four tones tests that are exceptionally clean. As soon as you amplify that to a watt or so, you’ll need to check the spectrum carefully. The drawback is they’re just not broadband enough, I see nothing that will match a 500kHz to 1500 MHz RTL-SDR. The closest thing I can find to that 144 to 1300 MHz discone as a TXDAC is the AD9122, a 1200 MHz TXDAC, DigiKey has at $68 in single piece quantities. This part still requires an external modulator.
Since the market for transmitters is always smaller than for receivers, I don’t see a prospect for a transmitter as broadband as the RTL-SDR. Virtually every service that can transmit requires a license which cuts down on the number of people who will want to get licensed, and every service that transmits is only allowed to use certain restricted pieces of spectrum. The places people can transmit without a license are things like the WiFi bands, the Family Radio Service, and the old Citizen's Band. All of these are small slices of frequencies, tiny pieces compared to the RTL-SDR bandwidth. Any place where a license holder can transmit wouldn't be allowed to use a "casual" transmitter like I'm describing; they'll have industry requirements their transmitter will have to meet, and they won't be authorized to make their own radios. Only hams can do that.
Many ham band only transmitters include very broad receivers; my Yaesu HT, an old VX-6R, includes the ability to receive from 500 kHz to 1 GHz (well, 998.990 MHz). I think we'll have to be happy with these. I don’t see a reasonably convenient way of putting together a transmitter to match the RTL-SDRs, and nothing even remotely capable of matching the cost/benefit ratio.