I performed a few simple PubMed searches -- the earliest publication that I can find specific to a Warburg-induced survival advantage is the attached 2002 paper.  Maybe there are earlier pubs that I missed.  This paper demonstrates a HIF1-dependent mechanism.  Since this paper there have obviously been additional mechanisms discovered. 

I performed a few simple PubMed searches -- the earliest publication that I can find specific to a Warburg-induced survival advantage is the attached 2002 paper.  Maybe there are earlier pubs that I missed.  This paper demonstrates a HIF1-dependent mechanism.  Since this paper there have obviously been additional mechanisms discovered. 

Probably you have already read it but there is a lot of information in the section of "reprogramming of energy metabolism" in the review in Cell by Hanahan and Weinberg in 2011. DOI 10.1016/j.cell.2011.02.013.

I haven't read the original Warburg article from 1930 but he may have implied it already at that time

The first mention (1926 not 1930) regarding glycolytic metabolism and tumors can be found in Warburg (Warburg O (1926) Ǘber den Stoffwechsel der Tumoren. Springer, Berlin). Glycolysis is inefficient in terms of ATP production, as it generates only two ATP molecules per molecule of glucose, whereas complete oxidation of one glucose molecule by oxidative phosphorylation generate up to 36 ATP molecules. Despite its low efficiency in ATP yield per molecule of glucose, aerobic glycolysis can generate more ATP than oxidative phosphorylation by producing ATP at a faster rate (Pfeiffer et al. 2001). Therefore, an inefficient but faster pathway for ATP production may be preferred to meet the high demands of dividing cells (Guppy et al. 1993). This mechanism is of strategic relevance under conditions of hypoxia or fluctuating oxygen availability (experienced especially during the early stage of cancer development) in which mitochondria cannot generate enough ATP; in those conditions aerobic glycolysis may give tumours a significant growth advantage (Postovit et al. 2002; Pouyssegur et al. 2006).

Glycolytic activity seems to correlate with the degree of tumour malignancy:  glycolysis is faster and oxidative phosphorylation is slower in highly undifferentiated and fast growing tumours than in slow-growing tumours or normal cells (Krieg et al. 2004). Furthermore, the fully transformed cell line is more dependent on glycolysis (and less dependent to oxidative metabolism) for ATP synthesis (Ramanathan et al. 2005). The tumour metabolome – namely the glycolytic phenotype – no doubt confers to the evolving cancer cell population a proliferative advantage and contributes to tissue invasion and metastatic spreading. However, such characteristics are not specific for cancer. Embryonic tissues (Seshagiri and Bavister 1991), as well as highly proliferating cells share a similar pattern, even if evolution has devised more than one metabolic solution to support cell proliferation (Valgepea et al. 2010; Xu et al. 1999). For a review on these arguments, see: M. Bizzarri et al. Bizzarri M, Dinicola S., Manetti C. Systems Biology approach to metabolomics in cancer studies. In: Systems Biology in Cancer Research and Drug Discovery. Azmi A. Ed. , Springer, 2012, ISBN 978-94-007-4818-7; Metabolism and cell shape in cancer: a fractal analysis. D'Anselmi F, Valerio M, Cucina A, Galli L, Proietti S, Dinicola S, Pasqualato A, Manetti C, Ricci G, Giuliani A, Bizzarri M. Int J Biochem Cell Biol. 2011 , 43(7):1052-8.

It appears that although glycolysis is vastly less efficient than mitochondrial ox phos, the fact that glycolysis can occur at a faster rate than ox  phos, provides the building blocks for nucleic acid synthesis, and that we see embryonic and other proliferating cells take the same metabolic route, are all arguments as to why the metabolic transformation seen in cancer cells provides advantages. While all of the above are rational arguments none of them prove this is why the warburg effect occurs and they remain hypothesis. At the end of the day the cause of the Warburg effect would appear to still be a matter of debate.

Mariano Bizzarri kudos!