My week in San Francisco, where I attended Salesforce’s annual AI conference and visited with the CEOs of Starbucks, Broadcom, and many others, was full of new insights about AI, stocks, and the market at large. Below are my top 10: Consider these tips to inform your investment decisions in the days and months ahead. 1. Salesforce’s AI platform is working – just not in the way some people think. Skepticism surrounding Salesforce’s AgentForce, its platform featuring agents and virtual robots to grow your business, is subsiding after CEO Marc Benioff’s keynote speech to kick off the company’s annual Dreamforce conference. It’s a two-pronged issue that has made Salesforce stock one of the worst performers in the market. One is that its Agentix initiative is so powerful that it has created geniuses who can write as good code as Salesforce’s legacy software unit produces; So, you no longer need legacy Salesforce. And second, because your agent codewriters are so effective, you can fire people off and so you won’t have to pay as much to Salesforce for its help as you currently do. Remember, businesses pay for legacy Salesforce by the “seat” or user, so if you have fewer sales people, you don’t need as much of the Salesforce product. Every year, Benioff gives an impressive 90-minute keynote speech about all of his new products with the aim of increasing sales for his customers. When I was helping run thestreet.com, I found its addresses very useful for getting new memberships. I learned a lot about what to do from keynote speeches. This time Mark’s presentation was very different. He brought in leaders from four companies – PepsiCo, Dell, FedEx and Williams-Sonoma – and showed how they were using agents to help them streamline their businesses, save money and increase revenue. Actual real use cases where no one cut back on legacy Salesforce and instead used it more efficiently. All customers had the option to cut down on inheritance – it’s now included as an option when you take Agentforce – but it doesn’t seem many are taking it. In fact, in an investor meeting during the conference, Salesforce raised its organic revenue growth rate from 10% for the 2026 to 2030 fiscal years — it fell to 9% — and said it expected sales of more than $60 billion in 2030, far exceeding analysts’ estimate of $58.37 billion. This estimate is important because the mark is beating the company’s internal estimates, but not the Street’s estimates. This became a continuation of disappointing quarterly progress. that’s over. Now he’ll show you that AgentForce is worth, hopefully in the billions of dollars over the next 18 months, and the stock will start climbing upward. I contacted three of the four CEOs who were part of the presentation to confirm what I wrote here, and I feel much better about my position than I have in a very long time. 2. OpenAI supporters are getting worried… Doubts about OpenAI’s spending are deepening as it moves toward consumer products like Sora, which generates realistic and animated videos from text prompts, and “erotica” for adults — both of which indicate it is not succeeding in the enterprise. Analysts hate consumer software because consumers are fickle and won’t pay for it. Enterprises pay too much and do not change vendors. The fact that OpenAI is already deeply ingrained among consumers has many of its supporters extremely upset. The company needs a huge revenue ramp next year to justify all its spending, and the crowd I sat with thinks that won’t be possible. 3. …but not Broadcom’s Hawk Tang. To counter that skepticism, I spoke with Broadcom’s very tough CEO Hock Tan, who has full confidence in the work we intend to do with OpenAI and says they will be a tremendous partner. Hawks are too tough and tough to work with pie-in-the-sky companies. It was difficult for me to reconcile my skepticism about OpenAI and Hawk’s strong belief in the company. Let’s just say it makes me more confident about OpenAI’s staying power. 4. AI buildout is not a zero-sum game. Lisa Su, CEO of Advanced Micro Devices, was the dragon slayer or David vs. Goliath on everyone’s lips for her battle with Jensen Huang and Nvidia. This is a false creation. These hyperscalers need everything they can get, I mean everything, and that includes custom chips from Broadcom, light-duty chips from AMD, and heavy-duty software-stuffed stacks from Nvidia. This is not zero-sum. Still people think that it is so. AMD’s chips won’t be ready until the middle of next year; By then, there will be a whole new iteration of Jensen, Vera Rubin, with dramatically more power than AMD’s offering. If you’re sold on Nvidia’s OpenAI deals with Broadcom or AMD, you’re making a mistake. But you can buy AMD and Broadcom to go with your Nvidia. 5. Now’s the time to sell your nuclear stockpile… We keep hearing that data centers use so much power that it’s going to replace the grid. I came away thinking that the grid is self-sustaining, but the energy it produces needs to be distributed better. Go back and listen to my interview with PG&E CEO Patty Pope. There is no shortage of power, he said, just a lot of power that can be generated when no one uses it. He said if you could spread the electricity use around and perhaps store it, that would be a better solution than what we are doing now. I felt better about the power part of the equation after talking to him and Hamid Moghaddam, CEO of Prologis, which has several large data centers powered by rooftop solar. Many technology executives don’t know much about electricity and should spend some time with large utility CEOs to better understand the problems on both sides of the equation. Furthermore, officials I spoke to said you would be crazy not to sell every uranium and nuclear stockpile. There is no real pressure to build new nuclear weapons, even small ones, or GE Vernova, the most established manufacturer, would be looking at orders. Sell every single one of these. Check out my new book, “How to Make Money in Any Market,” where I outline my system for building long-term wealth through your investments. 6. …your quantum stocks too. Quantum is not so ready for prime time that I would sell all this stock. No one is seriously talking about quantum as a way to get things done, and if so, it remains to be seen how IBM works with it commercially. Quantum stocks, like nuclear stocks, have been completely inflated and are living off the press releases of a particularly cash-strapped government. Although not pie in the sky, quantum computing breakthroughs will not produce commercial calculations for a very long time. All these companies need to make big equity offerings. Insider selling will soon take over, and they have really weak shareholders. 7. Don’t sleep on Meta’s smart glasses. Mark Zuckerberg’s vision of glasses and a tiny handheld pocket computer as a way to deliver AI to humans is taking shape faster than I thought. Meta’s glasses, which are a bit like Elvis-Costello, send huge amounts of information back to the cloud. But they also have a lot of computer power on campus, so to speak, because they’re loaded with Arm chips. These glasses can make calls, take photos, and speak multiple languages, among many other uses. It seems like they are a must for any traveler and any business person trying to work abroad. 8. If Dell’s stock goes down, you might want to buy it. Dell is moving away from all the other coordinators in the Nvidia installation and the racks that the chips are part of. I thought Hewlett Packard Enterprise was on the rise, but it had a miserable quarter. Nebius has real dreams, and I’ll sell him tomorrow; How far ahead is Dell? I regret that I did not take the initiative to buy Dell. If the share price goes down then it should be bought as it is performing much better than all the rest. I’m not buying into Bitcoin and data mining company IREN. Sell that and buy Corvea. This sale cannot be finalized immediately. But the insider selling that is about to come, and the corporate financing that should come, will end the upward spiral. 9. Levi Strauss is back. Away from tech, I was very impressed by the growth of Levi’s women’s line, and Levi Strauss is on a new growth path that is not reflected in the stock. My entire team was stunned by what CEO Michel Gass has offered and the stock is very cheap compared to the innovation he brought to the party. 10. San Francisco is coming back, too. The city of San Francisco is becoming safer thanks to the tireless work of Mayor Daniel Lurie. Density, which had disappeared, is coming back due to start-ups looking for cheap real estate. There are still lots of open storefronts and very few cops – hence Marc Benioff’s whimsical comments about the need for National Guard troops. There are fewer restaurants due to so many closures. Night time is still a bit quiet. But I felt safe. The dangerous tent cities and excessive drug use on the city streets have ended. It’s nascent, but real. (See here for a full list of the shares in Jim Cramer’s charitable trust.) As a subscriber to the CNBC Investing Club with Jim Cramer, you will receive a trade alert before Jim makes a trade. Jim waits 45 minutes after a trade alert is sent before buying or selling stocks in his charitable trust’s portfolio. If Jim has talked about a stock on CNBC TV, he waits 72 hours after the trade alert is issued before executing the trade. The above Investment Club information is subject to our disclaimer as well as our terms and conditions and privacy policy. No fiduciary obligation or duty exists, or is created, by virtue of your receipt of any information provided in connection with the Investment Club. No specific results or benefits are guaranteed.