https://en.wikipedia.org/wiki/List_of_natural_disasters_by_death_toll
It's the second in the section about the world's deadliest famines but it has no article and other references. It seems like a pretty big deal but there's no info on it online
First of all, the famine definitely happened. That you don't see a lot about it in English sources isn't too surprising though. Especially Wikipedia. Remember that Wikipedia is very America-centric, and things only get written if someone decides to take the time to write it.
So yes it happened, and it was one of the worst in a long time. Areas hit hardest were Jiangsu, Anhui and part of Henan. The cause of the famine was that there had been substantial amounts of rainfall leading up to it, which killed off a lot of crops in the area. These were reported to be the worst rains in 40 years. Hundreds of thousands of people were immediately affected by hunger, but many more died either as a result of violence or suppression of riots by the state. At this point the Qing was struggling to maintain control and stability, and the use of force to quell unrest was common. The riots began around March and by June, the military was sent to suppress them.
The worst of it was in 1907, but 1908 saw similar rains, and then in 1910 and 1911 famine struck again. These are often taken together, which is part of how you get the number as high as you see, but also the impact was really also just that severe, such that major relief organisation got involved, and people as far as America were raising money for disaster relief.
While the rains that caused the famine impacted East China the most, the resulting famine was felt throughout the country. More importantly, even when food production was recovering, the impact of the famine lasted for years to come, and was a major factor leading to the Xinhai Revolution which brought down the Qing.
1907 also saw the start of multiple armed uprisings by the Tongmenghui, the final one being the Wuchang Uprising which triggered the Xinhai Revolution. Displeasure with the Qing's handling of the disaster relief, combined with general dissatisfaction with the government in the first place, combined with overseas support in the form of disaster relief but also backing groups like the Tongmenghui, came together to end Dynastic China.
However as for the number, I'm not sure 25 million is correct if we're looking at a narrow window. I usually see somewhere between half a million and 2.5 million deaths for 1907. 2.5 million is quite reasonable. An American publication at the time gave the number of "nearly 3 million" who were on the verge of starvation, but mind you this is right at the start of it making international news.
25 million is not impossible, but I'd expect then we're looking at a large area and over more than just that year. Later publications (with sensational names) put the number at 24. The Encyclopedia of Disaster Relief puts it at 25, and a lot of others quote the same numbers without much sense of where that's coming from. The entire population of China at that time was around half a billion, and most were affected. The problem is, without knowing whether any given number is referring to all of China, or just the regions hardest hit, or just 1907, or the worst of the period (1907-1908) or everything through 1911, it's basically impossible to say.
I've done some digging to try to get better numbers, but difference sources have quite different opinions on that, and I don't have access to actual Chinese census numbers for the time. What you'd need to do is check the census for the entire range of years and geography that you'd want to calculate, and while someone's probably done that, I don't know what conclusions they came up with. Chinese sources have just as wide a range of numbers as English sources, so without actually going through reports from the time (which are not published in any way that I can access them at the present time).
I think for the time being we can say that the 25 million number is reasonable and certainly possible. If I dig up anything better I'll come back and update, but for now, yeah 25 is probably the upper end of an acceptable number.