Few companies have as strong a growth trajectory as Taiwan Semiconductor Manufacturing (TSM +0.55%). TSMC, as it's also known, is the world's largest contract chip manufacturer, which means that it acts as a fabrication company for those who do not have the capabilities to produce chips themselves. This positions TSMC nicely for the road ahead and makes it a must-own for nearly all investors.
However, a wrench was recently thrown into the AI investment thesis by DeepSeek's R1 model, which has taken the world by storm. So, is this a big deal for TSMC? Or will it be OK in the long run?
DeepSeek's AI model doesn't affect TSMC's trajectory
Although DeepSeek's R1 model is a true innovation, it doesn't affect the long-term viability of investing in TSMC. The biggest breakthrough DeepSeek found in its generative AI model is a way to make training more efficient, which requires less computing power.
The market assumed this meant that companies would need to buy less computing power to process these models, but that's not the right way to look at it.
NYSE: TSM
Key Data Points
Instead, AI spending will likely continue at the same pace, but the models will become more efficient. With more-efficient models and increased training power, the pace of innovation and usefulness will ramp up, accelerating the payoff period for all of these AI investments.
Regardless, this doesn't affect the TSMC investment thesis because it will benefit either by making chips used in GPUs to train these AI models or by making chips used in devices that run these AI models for an end user.
Because TSMC is a top supplier for nearly all sectors, it still has a bright future. So investors should consider taking advantage of the stock's dip; buying opportunities like this don't come around very often.
The stock is on sale after the drop
After nearly all AI stocks sold off on Monday following reports of DeepSeek's model, TSMC stock is about 10% off its all-time high. This represents great value, especially with the company trading for 29 times trailing earnings and 22 times forward earnings.
TSM PE ratio, data by YCharts; PE = price to earnings.
This price is a bargain for the growth that the chipmaker is expected to put up. Its management sees strong increases over the next five years, with revenue companywide expected to enjoy a 20% compound annual growth rate (CAGR). AI-related hardware is forecast to see even better gains, with management calling for a 45% CAGR for the next five years.
Chip orders are often placed years in advance, and a single breakthrough from an AI competitor will not change that pattern. The company's growth trajectory is still incredibly strong, and other factors will help it achieve that growth rate.
These include new chip technologies, which are slated to be launched later this year. TSMC's 2-nanometer (2nm) chip provides serious efficiency gains and can provide the same computing power as previous-generation 3nm chips, while consuming 20% to 30% less energy. Later on in 2026, the company's A16 chip will debut, providing an extra 15% to 20% energy savings from the 2nm chip.
Energy consumption is a huge operating cost for any AI hyperscaler (be it ChatGPT or DeepSeek), and so these new chip generations should be popular with hyperscalers.
Plenty of growth remains in the chip space alongside groundbreaking innovations. While DeepSeek's efficiency is impressive, it will likely be copied by many of the domestic AI platforms over the coming months and bring newfound efficiency to the AI world. This isn't going to stop AI investments, so the dip in TSMC's stock should be seen as a buying opportunity.
