The biosynthesis of human milk oligosaccharides (HMOs) using several microbial systems has garnered considerable interest for their value in pharmaceutics and food industries. 2'-Fucosyllactose (2'-FL), the most abundant oligosaccharide in HMOs, is usually produced using chemical synthesis with a complex and toxic process. Recombinant E. coli strains have been constructed by metabolic engineering strategies to produce 2'-FL, but the low stoichiometric yields (2'-FL/glucose or glycerol) are still far from meeting the requirements of industrial production. The sufficient carbon flux for 2'-FL biosynthesis is a major challenge. As such, it is of great significance for the construction of recombinant strains with a high stoichiometric yield. In the present study, we designed a 2'-FL biosynthesis pathway from fructose with a theoretical stoichiometric yield of 0.5 mol 2'-FL/mol fructose. The biosynthesis of 2'-FL involves five key enzymes: phosphomannomutase (ManB), mannose-1-phosphate guanylytransferase (ManC), GDP-D-mannose 4,6-dehydratase (Gmd), and GDP-L-fucose synthase (WcaG), and α-1,2-fucosyltransferase (FucT). Based on starting strain SG104, we constructed a series of metabolically engineered E. coli strains by deleting the key genes pfkA, pfkB and pgi, and replacing the original promoter of lacY. The co-expression systems for ManB, ManC, Gmd, WcaG, and FucT were optimized, and nine FucT enzymes were screened to improve the stoichiometric yields of 2'-FL. Furthermore, the gene gapA was regulated to further enhance 2'-FL production, and the highest stoichiometric yield (0.498 mol 2'-FL/mol fructose) was achieved by using recombinant strain RFL38 (SG104ΔpfkAΔpfkBΔpgi119-lacYΔwcaF::119-gmd-wcaG-manC-manB, 119-AGGAGGAGG-gapA, harboring plasmid P30). In the scaled-up reaction, 41.6 g/L (85.2 mM) 2'-FL was produced by a fed-batch bioconversion, corresponding to a stoichiometric yield of 0.482 mol 2'-FL/mol fructose and 0.986 mol 2'-FL/mol lactose. The biosynthesis of 2'-FL using recombinant E. coli from fructose was optimized by metabolic engineering strategies. This is the first time to realize the biological production of 2'-FL production from fructose with high stoichiometric yields. This study also provides an important reference to obtain a suitable distribution of carbon flux between 2'-FL synthesis and glycolysis.