The Personal Fat Threshold Theory (now proven by research) holds that each person has a unique, genetically determined capacity to store body fat.
Diabetes Mellitus is one of the oldest diseases known to medicine. It has been documented in history as far back as 400 BC.  Yet, as recently as 2000 AD we still had no concrete evidence on precisely what causes it or how to reverse it.
That isn't to say that we were completely ignorant to the general concepts. In general terms, obesity is a leading risk factor for type 2 diabetes.  However, not all obese people are diagnosed with type 2 diabetes - and some healthy-weight individuals are. While the risk increase is statistically powerful, genetics and other factors must play some role.
One of the newest theories on the root cause of type 2 diabetes first gained traction in 2007. The twin cycle hypothesis of the aetiology of type 2 diabetes establishes that excess fat in the liver and pancreas cause a vicious cycle of insulin resistance and decreased insulin production according to an individual's "Personal Fat Threshold" dictates at what point this process will trigger the spiral into type 2 diabetes, through the shutdown of insulin-producing beta cells in the pancreas.
The personal fat threshold concept was officially published in 2014
In their abstract from "Normal weight individuals who develop type 2 diabetes: the personal fat threshold", Roy Taylor and Rury R Holman described it as follows:
"Given the new evidence that early [type 2 diabetes or T2DM] can be reversed to normal glucose tolerance by substantial weight loss, it is important to explain why non-overweight people respond to this intervention as well as obese individuals. We hypothesize that each individual has a personal fat threshold (PFT) which, if exceeded, makes likely the development of T2DM. Subsequent weight loss to take the individual below their level of susceptibility should allow return to normal glucose control. Crucially, the hypothesized PFT is independent of BMI."
So the personal fat threshold varies by individual, and does not tie directly to BMI measurements. 
This makes perfect sense, as the widely accepted Body Mass Index has several inherent flaws. Highly muscular individuals for example, can often score as overweight - despite having an extremely low body fat percentage.
As far as we can tell, your personal fat threshold is largely genetically determined.
Proof of concept, and type 2 diabetes reversal finally established in 2020
Based on the twin cycle hypothesis of the aetiology of type 2 diabetes and the theory of personal fat threshold, Roy Taylor and other researchers set out to imperically prove that a reduction in body fat would reduce the symptoms of type 2 diabetes, regardless of the subject's weight. Using a 12 month study, they placed a test group of diabetics on an intense calorie-reduced diet. 
The results started to become apparent as early as the first week of the program. Within 7 days, insulin resistance in the liver was essentially gone. Over the next 8 weeks, pancreas function also began to improve significantly for most of the test subjects.
By the end of the 1 year study, the majority of recently diagnosed type 2 diabetics had completely recovered full insulin production, and were no longer displaying any diabetic symptoms. Those who had lived with diabetes for 10 years or more prior to taking part in the study were less likely to see full recovery, but some did.
The crucial factor here is that all subjects had significant weight loss due to reduced body fat (10% or significantly more in some cases), and a majority saw their symptoms either reduced or eliminated.
Additionally, this study also proved that beta cells in the pancreas were not being directly killed off as previously believed - but merely going dormant.
Follow-up research improves the outlook even more
This data was groundbreaking, but two issues remained. First, what were the long-term results for study subjects? And second, was a crash diet with massive weight loss absolutely necessary for success?
Five years after the initial study, researchers collected follow-up data from the participants. A number of the study subjects had regained some or most of the weight they lost during the study, and in most, type 2 diabetes returned with it. Those who maintained their weight loss however, were twice as likely to have no further symptoms at all.
After crunching the numbers, the research team determined that a 10% average body weight loss over a 5 year timeframe was sufficient to aleviate type 2 diabetes in a full 30% of individuals. 
Type 2 diabetes directly correllates with the personal fat threshold of the individual - which is largely dictated by genetic disposition. Put simply, it is the point at which insulin-producing beta cells in the pancreas begin to shut down due to stress caused by excess fat.
The ability of these beta cells to be revived and restored to function depends largely on the length of time the individual has lived with diabetes.
Significant weight loss (10% of body weight at minimum) is required in order to achieve remission - and as long as you maintain a body fat percentage that falls below your personal fat threshold, diabetes can be completely reversed indefinitely.
Your personal fat threshold does not change - so if you gain weight again and exceed the threshold, your type 2 diabetes will also resurface.