I'm Still Selling

Summary

This video covers Apple’s mixed earnings report, highlighting a significant revenue drop in China and Apple's slow progress in AI, which led to the speaker selling off Apple stock for a higher-growth portfolio. The second half discusses the hype around DeepSeek's AI model, which claimed to have trained for only $6 million, but the actual cost was likely much higher. Despite the excitement, the model doesn’t outperform Google’s Flash 2.0, and the hype around DeepSeek is deemed overblown. Ultimately, the video critiques misleading narratives in the tech market and stresses the importance of realistic expectations.

Takeaways

• 😀 Apple's earnings report showed a mixed performance with growth in revenue but issues in China, where revenue dropped 11%.
• 😀 Apple faces challenges in China, where its revenue missed expectations due to inventory problems and increasing local competition from companies like Huawei.
• 😀 Apple's slow progress in AI, particularly in the iPhone, is raising concerns as competitors have already integrated AI into their phones.
• 😀 Apple’s cautious and slow approach to AI is contrasted with the market’s growing demand for faster AI innovations.
• 😀 Despite Apple's stable growth, it does not meet the speaker’s portfolio criteria for higher growth investments, leading to a decision to sell Apple stock.
• 😀 DeepSeek’s AI model initially caused market uproar with claims it cost only $6 million to train, which was misleading.
• 😀 Investigations revealed that DeepSeek's actual cost to train its model was likely around $1.6 billion, significantly higher than the initial claim.
• 😀 DeepSeek’s V3 model was hailed as revolutionary, but comparisons show that it is not vastly superior to other models, such as Google’s Gemini 2.0.
• 😀 AI models, including DeepSeek, are becoming cheaper to run and more efficient over time due to algorithmic improvements and better compute power.
• 😀 Despite the initial hype around DeepSeek’s AI, its performance was not groundbreaking when compared to the latest models from companies like Google and Anthropic.

Q & A

What was the immediate investor reaction to Apple's earnings report?

Investors were initially enthusiastic, bidding the stock up by 3% after hours, but this was followed by a slowdown, and the stock's increase was reduced to only 0.5% by the next day.

What key issue affected Apple's revenue in China, and how did Apple address it?

Apple's revenue from China dropped by 11%, missing analysts' expectations. Tim Cook attributed more than half of this decline to inventory issues, raising concerns about whether Chinese consumers are favoring local brands like Huawei over Apple.

How does Apple's approach to AI compare to its competitors?

Apple has been slower than its competitors, like Google and local Chinese brands, in integrating AI into its devices. Their AI offerings have been met with a lukewarm response, and Apple has yet to roll out its AI features in China.

Why did the author sell $5,000 worth of Apple stock despite the company being profitable?

The author sold the stock because Apple’s growth rate is slower than what the author seeks in their portfolio, which aims for companies growing at a rate higher than the market average of 8%, preferably 12-15%.

What is the significance of Deep Seek's AI model release, and why did it cause market volatility?

Deep Seek's AI model generated a lot of attention due to claims that it was trained for only $6 million, which caused a market sell-off. However, further investigation revealed that the $6 million figure only covered a portion of the costs, and the actual cost of training the model was much higher.

How did Deep Seek's claim of training their AI model for $6 million hold up under scrutiny?

The claim was misleading. It was discovered that the $6 million figure only accounted for the GPU cost during pre-training, excluding other critical costs like R&D, employee compensation, and hardware. The true cost was estimated to be closer to $1.6 billion.

What performance claims about Deep Seek’s AI model were found to be overstated?

While Deep Seek's model was touted as groundbreaking and outperforming GPT-4, it was later found that when compared to newer models, such as Google's Gemini 2.0, Deep Seek's performance was not as impressive as initially claimed.

How does Deep Seek’s AI model compare to models from US companies like Google and OpenAI?

When compared to the latest models from US companies, such as Google’s Gemini 2.0, Deep Seek’s AI model is not as superior as initially reported. Google’s model is cheaper and offers similar or even better performance in some areas.

What does the author suggest about the future of AI development costs and model performance?

The author predicts that AI development costs will continue to decrease over time. Smaller models will achieve comparable performance to larger ones, and costs could fall further in 2025, making the achievements of Deep Seek and other companies more common.

What was the primary cause of the market sell-off following the Deep Seek news?

The primary cause of the market sell-off was the misleading $6 million claim about the cost to train Deep Seek’s AI model, which led to speculative hype and a subsequent market reaction based on incorrect information.