Optical interconnect is a fundamental requisite to realize Internet-scale data centers due to capabilities and benefits of optical devices. Optical interconnects are energy efficient and offer massive bandwidth support. State of the art interconnects can be divided into three types based on the optical technology used: 1) micro-electromechanical system (MEMS) optical cross connects (OXCs), 2) arrayed waveguide grating routers (AWGRs) and 3) semiconductor optical amplifiers (SOAs). MEMS switches are based on mature technology, have low insertion loss and cross-talk, and are data rate independent. They are also the most scalable and the cheapest class of optical switches. However, the reconfiguration time of these switches is in the order of tens of milliseconds. An AWGR switch is a passive device and works in conjunction with tunable wavelength converters (TWCs) or tunable lasers (TLs) while an SOA works as a gate element that manipulates light and also compensates for losses that occur during transmission of optical signals. AWGR and SOA switches have switching time in the range of nanoseconds but they are expensive as compared to MEMS. In this paper, we propose a novel all optical core interconnection scheme that utilizes potentials of both slow and fast optical switches. The core idea is to route traffic through slow or fast optical switch so that minimum end-to- end latency is achieved. Our architecture employs a single stage topology which allows our design to both incrementally scaled up (in capacity) and scaled out (in the number of racks) without requiring major re-cabling and network reconfiguration. We evaluate performance of the system using simulation and investigate a trade-off between cost and power consumption by comparing it with other well known interconnects. Our technique demonstrates a considerable improvement in power consumption and low latency with high throughput is achieved.