I'd say there is no single reason but a combination - plus a certain PoV.
There Is Only the TI-84plus
Not really. There are for one multiple models for different use cases from TI as well as other manufacturers, most notably Casio, but also HP (HP-39 GII, HP Prime) and Sharp (EL-9900G). In fact, TI was the new kid on the block with its TI-81, 5 years after Casio's ground breaking fx-7000G and as well years after Sharp and HP.
TI has always been more of a follower than a leader in development of these. The overwhelming rule of TI-84 is a strictly US phenomena, in other places of the world it's rather Casio. Especially when CAS variants are allowed.
TI Did Not Advance Since the 84plus
TI did. They maintained different series and models within all the way from 89 and 92 to Voyager and Nspire models. The latest of them, the Nspire CX II being introduced in 2019. It features actual ARM CPUs, huge memory, touchscreen and modern (*1) languages like Lua and Python.
But Development is Slow, Right?
Kinda. While development is constant, it doesn't run at overdrive at any of he manufacturers. Several reasons may be named:
It's a Market Cornered by a Duopoly
TI and Casio essentially own the market. HP and Sharp have only limited offerings. Like with every market where the overwhelming sales numbers are with two manufacturers, incentives are small to add features - that is past improvements in manufacturing to reduce cost. Even more in the US where Casio is in most states a distant number 2 if allowed at all.
It's a Niche Market
Unlike programmable calculators of the 1970s/80s, these modern calculators are no longer a business tool for engineers or in finance. Back then, those were the only way to get the needed power and flexibility on an engineer's desk - at an affordable price. But since the mid 1980s, maximum 1990, a PC can be considered standard in such positions, offering way more flexibility and capacity and work flow integration than any pocket calculator.
Graphics calculators developed right after that turning point and with a special use case: Teaching. They enable students to see the result of certain equations right away without the need of calculating it by hand and plotting it point for point on paper.
It's much the same as with (seemingly) basic desktop calculators being still manufactured, bought mostly by accounting/trades people with a solemnly need to summing up a series of item price times items sold.
It's a Situational Market
This goes with the niche part, but in time wise. Owners only need those calculators only for a rather limited time with a defined (and hopefully happy) ending. A special short-term phase in life, after which they do not really add above any other programmable. And be of way less usability than any PC software for the same purpose.
There is no need to buy one outside that phase - that is unless one stays in school as a teacher.
It's a Mature Market
The technology used already fulfills the purpose they are intended for. Neither faster CPUs, more storage or higher display resolution will make much difference. It's a bit like the PC since about 2010, when CPU and GPU speed became more than sufficient for 99% of all business and 90% of all home application. A situation the fast-moving phone market seems to reach soon - if it hasn't already.
It's a Regulated Market
Buyers are not free to choose manufacturer/model for features they like, or consider advanced, but are restricted to a defined feature set, set by authorities like teacher, school board or state regulations. The selection is usually even more restricted to certain models to simplify conformity checks by teachers.
Casio's Class-Pad series gives a good example. They were, all the way to the most beautiful fx-CP400, not allowed for most US tests (SAT etc.) as they could bring up a virtual QWERTY/Z keyboard. Casio tried to overcome this by special ROM versions disabling features, like they did for other countries, but to no avail. Finally they introduced in 2017 the US-only fx-GC500 disabling all keyboards except for an alphabetical sorted one. Notably it also features a different case colour (grey instead of black) so teachers can see the difference.
This is BTW not only a US issue, but worldwide. For example the TI-Nspire CX models do feature an interface for an optional wireless module, while European (CX-T) do not.
Without freedom to choose for features, there isn't much incentive for manufacturers to compete with additional features.
They Do Not Drop in Price
Yes and no. Sure, the nominal prices didn't drop much, but using inflation rate (and that calculator of yours) will show that 120 USD in 2004 is roughly the same as ~195 USD today (2023). So being sold today new at 150 USD means a 25% price reduction. Sounds more ok, doesn't it?
Also, every product (segment) has a low price below which selling will get impossible. We often think of price mostly related to production cost, but that is only true in theory and big numbers. In real-world scenarios every transaction has minimum cost and minimum turn around to be operational and profitable. It's why even at Walmart a cheap-ass 4-function calculator is still around 10 USD and a TI-30 with way more functions about the same price.
Price does not depend on function delivered, but defines functions possible. No matter if low end or higher up. At the 10 USD level it doesn't matter what functions are provided as they all come at the same price of a dirt-cheap mass-produced chip. The 10 USD price point was reached already 30 years ago, and stayed there since then. In fact, it's the same with PCs or tablets. Already since years they rarely go below 300 USD for a usable configuration.
Similar the price point for a graphics calculator made to fit school needs. Unlike that 10 USD TI-30, which didn't change since the mid 1980s, they do need updates and maintenance to stay within changing scope. And they need more support - not to mention the 'marketing' needed to place them at schools.
Why Not Using a PC Or Tablet?
I guess we can agree that a PC on every students desk isn't exactly a good idea. Has been tried, didn't work. Not only for size but as well the issue of making sure all pupils use the same unmodified software. Tablets may solve the size issue, less the software issue. Not every school is able to afford highly trained IT-capable teachers :))
But we'll get there ... slowly but we do, which at some point may make graphic calculators surplus.
*1 - Well, modern in sense of fashionable - as if the good old BASIC wasn't any good for those youngsters?