import pandas as pd
import numpy as np

# Eigentlich:
# df = pd.read_excel .... 
# Vorsicht mit der deutschen Punktation beim Lesen!

df = pd.DataFrame(['A',100,105,45,'B',700,500000,23,247,5047,'C',1234,5678,'D','582109','E','F'],columns=['data'])

Qnumber = df['data'].str.fullmatch(r'[0-9]+',na=True)

df['group'] = np.nan
df['numeric'] = np.nan
df['sum'] = ''

df.loc[~Qnumber,'group']=df.loc[~Qnumber,'data']
df.loc[~Qnumber,'sum']=df.loc[~Qnumber,'data']
df['group'] = df['group'].ffill() # fills np.nan with last letter

df.loc[Qnumber,'numeric'] = df.loc[Qnumber,'data'].astype(int)

sumdic = df.groupby('group')['numeric'].sum()
df.loc[~Qnumber,'sum'] = df.loc[~Qnumber,'sum'].apply(lambda x: str(int(sumdic[str(x)])))

# Variante, in der das nach den Buchstaben in die gleiche Spalte geschrieben wird:
# df.loc[~Qnumber,'data'] = df.loc[~Qnumber,'data'].apply(lambda x: str(x) + ' ' + str(int(sumdic[str(x)])))

result = df[['data','sum']]

print(result)

# dann noch result.to_excel ...

data = df["data"]
pd.to_numeric(data, errors="coerce").groupby(data.where(data.str.isalpha()).ffill()).sum()

Out[50]:
data
A       250.0
B    506017.0
C      6912.0
D    582109.0
E         0.0
F         0.0
Name: data, dtype: float64