To acquire competitive advantages in an ever-changing business environment, many business gurus step up to outsource their non-core activities to third-party members. Except for some basic activities such as operations and marketing, R&D outsourcing has become prominent since 1970s. There are many success examples such as IDEO, an international design and consulting firm founded in California, in 1991, participating in the design of Apple Mouse. Different from other activities, R&D outsourcing is with high uncertainty and information asymmetry. The risk sharing and information issues are very important for both participants. To tackle with these problems, a principal-agent like R&D outsourcing problem between a technology provider and buyer is studied. Since the buyer can't observe the provider's R&D effort, an optimal contract must be designed to alleviate the moral hazard problem. The contract is composed of two parts, namely fixed payment and unit payment. Fixed payment mean no matter what the random output is, the provider will get this part. The unit payment is the gain for each unit output. Faced directly to R&D risks, the technology supplier is regarded as a downside risk averse agent. The provider's risk aversion attitude is modeled with Value-at-Risk (VaR) constraint. By solving this optimization problem with KKT conditions, the VaR results are compared against the seminal mean-variance framework, and it is founol that: 1) the unit payment under VaRincreases as the demand volatility increases, while it decreases under M-V framework; 2) as the risk aversion increases, unit payment in VaR increases while M-V decreases; 3) as the provider becomes more risk averse, the fixed payment under VaR increases; 4) reserved profit and risk control level plays different roles in determine the optimal parameters, the reserved profit doesn't impact on the unit payment while the risk control level influences them both. VaR method not only overcomes the shorting comings of mean-variance method in theory, such as M-V can't separate upper gains with downside loss. The VaR method also provides much more clear managerial insights for technology outsourcing problem.