蜜豆视频

Video:听CIO鈥檚 Sundeep Gantori: DeepSeek and its impact on the AI value chain (1:46)

Podcast:Investors Club 鈥� DeepSeek's emergence, tech sell-off, and big tech earnings in focus (10:26)

Big tech earnings reports in recent days have offered investors more insights into the AI outlook in a turbulent week for the sector. Tech stocks sold off at the start of the week amid concerns that a new open-source AI model from Chinese startup DeepSeek could affect the profitability of US tech leaders.

On Wednesday, Facebook-parent Meta reported strong fourth-quarter revenues and gave a bullish capex outlook. Microsoft also beat quarterly revenue estimates, though its shares fell sharply on Thursday after it issued mixed guidance for its Azure cloud business, which in 4Q was dragged down by weaker-than-expected non-AI businesses. On Thursday, Apple hit all-time records for total company revenue and earnings per share (EPS) and beat consensus estimates for its fiscal 2025 first quarter. The company reported revenue of USD 124.30 billion, slightly above consensus estimates of USD 124.12bn, and EPS of USD 2.40 relative to expectations of USD 2.35.

Tech sector volatility is likely to continue in the months ahead. But we see the initial set of large-cap tech results as reassuring and believe the AI growth story remains intact:

The emergence of lower-cost, lagging-edge models is not surprising, in our view. We estimate that, on average, the cost of training a large language model should drop by 75% in a year鈥攁nd DeepSeek鈥檚 version 3 model is in line with this cost curve evolution. In other words, the ability of DeepSeek to replicate an older model at a lower cost should not be surprising. And on inference, we do not expect DeepSeek鈥檚 R1 to prove significantly cheaper than most comparable models. So, while DeepSeek indicates that China is likely to remain a strong force in AI innovation, we do not believe that its emergence materially changes the growth outlook for the industry or key parts of the value chain. Additionally, we believe that鈥攍ike in many industries鈥攁 mix of low-cost and high-cost solutions can coexist, serving different demands and use cases.

Big 4 capex plans remain solid despite lower-cost models like DeepSeek鈥檚. With a strong AI outlook from both Microsoft and Meta, we still expect the combined capital spending by the Big 4 tech companies (Alphabet, Amazon, Meta, Microsoft) to rise 25% this year to USD 280bn鈥攁 positive for the AI compute industry. Big tech firms are likely to continue to spend on higher-cost, cutting-edge frontier models in their pursuit of artificial general intelligence (AGI). Meanwhile, we also expect broadening spending from tier-2 platforms, businesses, and sovereign players to help keep overall AI capex solid.

AI adoption trends are improving, which bodes well for cloud growth. The recent Census Bureau survey shows AI adoption rates in the US rose to 6.3% in the fourth quarter of 2024 from 5.9% in the prior quarter. More importantly, with multiple reasoning models likely to be introduced in the next few months, the department鈥檚 survey suggests AI adoption rates in the US are expected to improve to 9.2% by the middle of 2025. Growing adoption rates should boost AI monetization trends, helping narrow the gap between AI capex and revenues for big tech companies. In particular, we believe rising AI adoption should drive strong cloud growth.

Putting all this together, we continue to see a favorable risk-reward for our AI theme. As we discussed in our CIO Alert, AI is here to stay鈥攁nd if anything, DeepSeek reinforces that. Notwithstanding some currency headwinds from the strong US dollar, we maintain our 2025 earnings growth forecast for global tech at 18% and expect AI to remain the main driver for the sector this year. That said, with the AI rally now more mature, investors should be prepared for further volatility, as any uncertainties around interest rates, tariffs, regulations, or company-specific factors (like product cycles) could lead to short-term pullbacks.

Without taking any single-name views, we believe investors can take advantage of heightened volatility鈥攆or example, through structured strategies鈥攁nd build positions for longer-term growth in the AI ecosystem (e.g., through our AI TRIO theme), as we believe the focus will ultimately return to robust fundamentals. As value can shift quickly within the AI ecosystem, investment approaches that are too concentrated or overly passive can be risky, and we favor an active and diversified approach to navigate the enabling, intelligence, and application layers of the AI value chain. We also remain constructive on China鈥檚 tech sector, which could be a particular beneficiary of DeepSeek鈥檚 open-source model, given the increased trade and technological barriers between the US and China.

More broadly, we expect the greater efficiency from new, lower-cost algorithms to lead to increased economic productivity, which is supportive of the broader equity market. In addition to these potential productivity gains, we believe the combination of solid US economic activity, healthy earnings growth, lower borrowing costs, and the potential for greater capital market activity will lead stocks higher over the balance of 2025. In our base case, we see the S&P 500 reaching 6,600 by the end of the year.

To read more, please see 鈥�Investment conclusions post DeepSeek鈥檚 R1 release鈥� and 鈥淚ntelligence weekly #20: Staying positive on AI after initial results and evaluating bear and bull cases.鈥�