Artificial Intelligence

​Don't let the recent downturn in the AI trade mislead you. The reality is that AI-powered service providers are struggling to keep up with demand, and AI infrastructure providers are unable to build capacity fast enough. All the potential that drove AI stocks higher in 2024 remains intact. At Deepwater, we continue to believe there are still 2–3 years left in this bull market.​

Deepwater’s second annual Frontier Tech Survey measures the American public’s viewpoint on emerging technologies, including AI, Electrification & Automation, the Creator Economy, and Relationship with Tech. Key Takeaways: AI – General usage is increasing but we are still early in the…

All four key hyperscalers guided CapEx spending in 2025 to grow at a faster rate than previously anticipated before the DeepSeek scare. This signals that the most advanced AI companies in the world believe the path to AGI will demand an obscene amount of hardware. Yet, despite this clear trajectory, many AI hardware companies are still struggling to recover from the losses triggered by DeepSeek.

What Does DeepSeek Mean for AI? Not as much as the market would have you believe. This past Monday, most publicly traded AI companies saw a sharp devaluation from the emergence of a new AI model called DeepSeek R1. DeepSeek…

The AI trade still has room to grow as the spotlight gradually shifts toward software, underscored by investors' reactions to Salesforce's comments following their most recent quarter. The bottom line is I believe that while software companies have yet to see significant AI-driven revenue acceleration, investors are increasingly confident that day will come. This is good news for my view that AI will continue to be the catalyst driving the market higher, until it eventually bursts.

The Nasdaq was up about 5% this week following the election, versus the Mag 7 up 7%. Below I outline what I believe to be the reasons for the move and what it means for tech investors over the next…

Meta and Google are shifting their development focus away from Virtual Reality (VR) and Mixed Reality (MR) to ambient computing in the form of smart glasses given the potential that artificial intelligence adds to the form factor. I believe the annual market for smart glasses will reach hundreds of millions of units over the next decade, leading Apple to refocus its Spatial Computing initiatives to be more in line with Meta and Google wearables.

We asked Gemini, ChatGPT, Claude, and Grok 15 questions about upcoming events to determine which models are most effective at looking into the future. This note is centered on “predictions” in the balance of 2024, which we’ll cover as they…

We tested ChatGPT, Gemini, Grok and Claude to determine which models provide the most clear and accurate responses. Our theory is not all foundation models are created equal. The two takeaways from our first benchmarking are: first, there is no single best model; certain models are better at handling specific queries. Second, there generally are two model "personalities"; with Grok as the only model with a distinct "personality". As for the what does this mean for search question: the answer is Google will win. What is most important for the future of ChatGPT, Grok, and Claude is their ability to allow third parties to create unique, deep, and conversational experiences that deliver insights beyond search. That doesn't exist today.

Apple made good on expectations, announcing generative AI integration across its software and applications. Accessing these features will require 85% of Apple users to upgrade some or all of their devices over the next few years. This means revenue growth in FY25 and FY26 will likely be +10%, ahead of the current 6% and 4% expectations, which should be a positive for AAPL.

Shares of Meta traded down 11% on the March quarter guidance that the company is entering into a multi-year investment cycle starting with AI capex in 2024 tracking to be 12% higher than previous expectations. We believe investors are missing the point. Meta has a golden opportunity in the next decade to leverage its data and build AI tools and products for billions of users, advertisers, and businesses.

  We may be approaching the day when a $200M Pixar-quality movie, which typically takes four years to make, will be created in six months for $20K (aka, 0.01% of the budget and less than 10% of the time.) I…