本研究以台灣IC設計產業為研究對象，探討人工智慧（Artificial Intelligence， AI）技術快速發展之背景下，半導體企業之估值水準與籌資成本之間的關聯性。自2022年生成式AI與大型語言模型興起以來，全球市場對高效能運算（High Performance Computing， HPC）、AI伺服器及先進製程晶片之需求快速攀升，帶動半導體產業營收結構與資本支出出現明顯轉變，資本市場亦重新評價具備AI供應鏈優勢之企業。 本研究選取六家台灣上市櫃IC設計公司為樣本，並依其產品定位與AI業務布局區分為AI組（聯發科、創意電子、世芯-KY）與非AI組（聯詠科技、瑞昱半導體、神盾科技）。研究方法以比較分析法為主，分別採用Tobin′s Q衡量企業估值，採用資本資產定價模型（Capital Asset Pricing Model， CAPM）估算股權成本，並進一步結合公司資本結構、稅後負債成本，計算各公司2021年至2025年之加權平均資本成本（Weighted Average Cost of Capital， WACC），藉此比較兩組企業於估值與資金成本面之差異及其成因。 實證結果顯示，AI組企業之平均Tobin’s Q為6.42，明顯高於非AI組之2.01，反映市場對AI相關企業未來成長潛力給予較高評價。在資金成本方面，AI組企業五年平均股權成本為10.76%、平均WACC為10.23%，皆高於非AI組之8.77%與8.31%。此一現象顯示AI相關企業雖享有較高市場估值，惟因其系統性風險（Beta）較高，投資人要求之風險溢酬亦相對較高，導致其資金成本同步上升。然而市場仍願意給予AI企業較高估值，顯示投資人認為其未來成長機會足以補償所承擔之較高風險。 整體而言，AI浪潮已成為影響半導體企業估值與籌資成本之重要因素。本研究結果除可作為投資人評估半導體企業價值之參考外，亦可作為企業未來資本支出與財務規劃之依據。;This study examines the relationship between corporate valuation and the cost of capital of semiconductor firms under the rapid advancement of artificial intelligence (AI)， focusing on Taiwan′s integrated circuit (IC) design industry. Since the rise of generative AI and large language models in 2022， global demand for high-performance computing (HPC)， AI servers， and advanced-process chips has surged， reshaping the revenue structure and capital expenditure of the semiconductor industry， and prompting capital markets to re-price firms with AI supply-chain advantages. Six listed Taiwanese IC design companies are selected as samples and classified into an AI group (MediaTek， Global Unichip， Alchip) and a non-AI group (Novatek， Realtek， Egis) according to their product positioning and AI business exposure. Using a comparative analysis approach， the study measures corporate valuation with Tobin′s Q， estimates the cost of equity with the Capital Asset Pricing Model (CAPM)， and computes the Weighted Average Cost of Capital (WACC) for each firm from 2021 to 2025 by incorporating capital structure and the after-tax cost of debt. The empirical results show that the AI group exhibits a markedly higher average Tobin′s Q (6.42) than the non-AI group (2.01). In terms of capital cost， the AI group reports a five-year average cost of equity of 10.76% and an average WACC of 10.23%， both higher than the non-AI group′s 8.77% and 8.31%. These findings indicate that， although AI-related firms enjoy higher market valuations， their higher systematic risk (Beta) raises the risk premium demanded by investors， thereby increasing their cost of capital. Nonetheless， the market still assigns higher valuations to AI firms， suggesting that investors believe their growth opportunities are sufficient to compensate for the additional risk. Overall， the AI wave has become a significant factor influencing the valuation and cost of capital of semiconductor firms. The results provide a reference for investors in valuing semiconductor companies and for corporate capital-expenditure and financial planning.
