Metformin-related lactic acidosis is mainly due to the inhibition of hepatic gluconeogenesis.
Naturally, this leads to a buildup of pyruvate in the liver, and therefore increased available substrate for lactate production. However, this is not the only mechanism (after all, the pyruvate could just as easily be metabolised in the Krebs Cycle).There are several other factors which contribute, and one specific article which addresses the pathogenesis of biguanide lactic acidosis with a satisfying amount of detail.
At higher concentrations, metformin can inhibit the electron transport chain of the mitochondria, producing elevated levels of intracellular NADH. This reduces flow through Krebs cycle, so less pyruvate is cleared in this way. On top of that, the NAHDH:NAD ratio in the cytosol increases in favour of NADH, which favours the conversion of pyruvate to lactate.
Now, add to this a relative insulin deficiency (which reduces the activity of pyruvate dehydrogenase) and proper lactic acidosis ensues. Insulin deficiency accelerates the presentation of pyruvate precursors to the liver (these are mainly amino acids which get converted to pyruvate somewhere along the process of being converted into glucose).
Thus, we have an excessive buildup of pyruvate, with no hope of turning into glucose by gluconeogenesis, and with decreased entry into Krebs cycle. The only way to metabolise it is by LDH, into lactate. Thus, lactic acidosis ensues.
Being cleared renally, metformin therapy relies on normal kidney function. The patient with renal impairment will chronically retain metformin, and for them the dose is lower. As to the exact amount? It seems even 63 grams is survivable (albeit with hemodialysis and having presented with a lactate of 17.7 mmol/L)