Victorian style homes and orange barricades along narrow residential street during utility construction, in the Hayes Valley neighborhood of San Francisco, known locally as Cerebral Valley due to the large number of artificial intelligence industry workers present in the area, August 19, 2025. (Photo by Smith Collection/Gado/Getty Images)
Gado via Getty Images
Nvidia stock will move the market on Wednesday after the close, CNBC’s Jim Cramer predicted.
That’s because the AI chip designer will deliver a third-quarter earnings report on Nov. 19, when it could shake the markets up — or down.
If Nvidia beats expectations, raises guidance, maintains its lead over AMD, and says the company’s new Vera Rubin line of chips is ready to go, $NVDA could rise. Otherwise, look out below, Cramer warned.
Along with OpenAI, Nvidia is the co-leader of the booming generative AI industry. Without the continued success of these two AI juggernauts, the entire industry value network of chip manufacturers, AI cloud services providers, large language model builders, AI tool makers, and AI application developers I wrote about in Brain Rush will also suffer a drop in value.
While there is no question we are in an AI bubble, I wrote in October that I see the opportunities for investors exceeding the risks — for now. At the Cerebral Valley AI Summit in San Francisco last week, I learned meaningful insights that largely reinforced that conclusion.
Along with mixed views on the prospects for specific AI companies and some eccentric personalities, I observed a diverse collection of knowledgeable, passionate, articulate, and savvy investors and entrepreneurs aiming to create valuable products for their customers.
Key AI Investor Take-Aways
Investment opportunities and threats: Two venture capitalists agreed that AI is in a bubble, but they said that is where you can make money. They detailed some of their investments. Also an anonymous survey of 300 summit attendees revealed a bullish view for Nvidia stock and a pessimistic outlook for OpenAI.Fast-growing AI business models: Using vibe coding to create new businesses quickly — notably Magic School that lets teachers use AI to create and grade assignments — is a new business model. Other interesting examples include using AI to make the legal industry more effective and replacing human customer-service employees with AI-powered voice. Many compelling business leaders: This conference featured many fascinating founders, including the co-founder of Instagram and Anthropic’s current chief product officer, Mike Krieger, and many others leaders with deep knowledge of their problem spaces who struck me as extremely thoughtful and articulate.Social and political oddities: San Francisco Mayor Daniel Lurie – who proudly pronounced crime down 30% in the city – reluctantly responded to a question about how he kept the federal government from sending border agents to Fog City. Moreover, many of the founders struck me as Musk-pilled by the idea that smart people should have lots of children.
Here’s what to expect from Nvidia’s third-quarter report and how these observations could shape generative AI’s future.
What To Expect From Nvidia’s Q3 Report
Expectations are high for Nvidia’s third-quarter report. That’s due in part to the company’s track record. The AI chip designer, “has beaten Wall Street earnings estimates 90% of the time over the past five years but averaged only a 6.5% beat in the last four quarters,” according to Yahoo! Finance.
Investors are looking for revenue and adjusted earnings per share for Nvidia to increase 56% and 54%, respectively, to $54.83 billion in revenue and $1.25 in adjusted EPS, Yahoo! Finance reported. That’s about the same as the company’s second-quarter revenue growth of 55.6% and more than the 59% EPS increase last quarter.
Ongoing restrictions on Nvidia’s ability to sell advanced H20 AI chips to China could “shave several billion off potential revenue,” wrote Yahoo! Finance. Nvidia CEO Jensen Huang recently confirmed the likelihood this situation will not change.
Record-beating data center revenue coupled with higher profitability resulting from “economies of scale and software integrations, such as CUDA,” could yield a better-than-expected report, concluded Yahoo! Finance.
AI Investment Opportunities And Threats
Given Nvidia’s dominance of the AI chip industry, the future for investors depends on whether there is an AI bubble blowing up, how soon it will burst and how severe the damage will be.
I suggested last month two more optimistic and likely scenarios:
Companies discover and deploy applications of AI that generate enough incremental growth and profit to make the estimated $3 trillion investment in AI pay off. AI investment continues at a slower pace and valuations adjust accordingly.
The more than 300 industry insiders attending the Cerebral Valley AI Summit represented some $500 billion in value, the conference organizer Eric Newcomer said. Two venture capitalists there said that there was indeed a bubble and saw an opportunity.
“Yes, there is an AI bubble,” admitted Kleiner Perkins partner Ilya Fushman and investor Elad Gil. Bubbles are good times for investing and while many companies are funded, only a few survive the burst, Gil said.
Such bubbles are nothing new – they occurred in the auto industry when it was taking off. During the dot-com boom, Gil said thousands of companies went public, a few dozen survived, and two giant companies — Amazon and Google — remain.
Where is the investment opportunity in the AI stack? Foundation models have the potential to get five to 10 times bigger, Fushman and Gil said.
Gil sees opportunity in AI infrastructure. He invested in Brain Co. (along with President Trump’s son-in-law Jared Kushner), which has developed tools that use AI to “help hotels streamline bookings, enable governments to process construction permits, and assist insurers in managing claims more efficiently,” Forbes reported.
Gil also invested in Long Lake, a real estate roll-up startup focused on consolidating homeowners’ associations.
Fushman has backed Harmonic, which provides “mathematical super intelligence,” and Profound, a platform for measuring brand visibility. Both Fushman and Gil have backed Harvey, which helps make lawyers more productive in their corporate work and has reached $100 million in annual recurring revenue.
Another interesting investment angle is the results of an anonymous survey of the 300+ attendees at the conference. They estimated Nvidia would be worth $6 trillion in 2026 and OpenAI will hit $30 billion in 2026 revenue, but more attendees want to invest in Anthropic.
The most popular unicorn to short is Perplexity, which is valued at $20 billion, according to PitchBook, which I used as the source of all valuations in this article. Second most likely to drop is Open AI ($500 billion) and several companies were in third place, including Cursor ($29.3 billion), Mistral ($13.7 billion) and Thinking Machines ($10 billion).
Fast-Growing AI Business Models
There were many interesting AI business models. I was most interested in:
How vibe coding is spinning up businesses with enough revenue to be valued at $500 million very quickly; The aforementioned Harvey (valued at $9.33 billion), which uses AI to enable law firms to grow more profitably; and ElevenLabs ($3.3 billion), a leader in voice AI replacing customer service with AI while enabling celebrities like Matthew McConaughey and Michael Caine to monetize their voices, Variety reported.
The vibe coding startups use Replit ($3 billion). Its CEO Amjad Masad cited Magic School which helps teachers create and grade assignments. Founded in late 2023 it is valued at $452 million.
I am sorry Masad declined to name the JP Morgan executive who used Replit to create an investment-banking service that generated $1 million in contract revenue in just weeks; Masad said it’s valued at $500 million.
Such business models offer a way for domain experts to turn their insights into products that generate significant value for customers. In general, such startups will not survive if they are competing on price — they must make customers better off, Masad concluded.
Compelling AI Business Leaders
There were several speakers — mostly founders — who struck me as especially brilliant and compelling. These include Andy Konwinski, who sports a very long beard and who co-founded Perplexity, Databricks, and Laude Ventures. He is investing $150 million from the most highly-paid AI scientists to invest in their projects.
Before starting Harvey, its CEO Winston Weinberg was an associate at O’Melveny & Myers in antitrust and securities litigation. Harvey has grown to 440 people and includes many domain experts, which strikes me as interesting.
How so? Decades ago I was an early employee in Applied Expert Systems, an MIT expert-systems startup that interviewed domain experts and coded their decision rules for personal financial planning. Like ApEx, Harvey is tapping into the value of human subject matter experts, albeit far more successfully.
The company has 800 customers, mostly law firms that use the company’s AI agents to serve their corporate clients more efficiently by getting the right data and using the right tools, Weinberg explained. Here is an excellent TechCrunch interview on Harvey’s entrepreneurial journey.
Mati Staniszewski, CEO of ElevenLabs, is another fascinating founder. “Its technology creates human-like AI voices, converts text to speech and dubs audio in 29 languages. With partnerships spanning The Atlantic, Washington Post and TheSoul Publishing (known for 5-Minute Crafts),” according to Forbes.
Christina Cacioppo CEO of Vanta ($4.2 billion) was a great communicator. Founded in 2018 and based in San Francisco, the 1,411-employee company develops a compliance management platform that automates risk management, security audits, and vendor assessments. Cacioppo has a deep knowledge of risk management and security and flings neologisms like “AI-pilled” to refer to employees who have grown up with AI.
Finally, Krieger, the chief product officer at Anthropic ($183 billion), described with eloquence the company’s design challenges, such as where Claude should be on the border between empathy and sycophancy towards users.
Social And Political Oddities
The conference concluded with Mayor Lurie and Jimmy Ba, a professor at University of Toronto who cofounded xAI.
Lurie talked about how crime has dropped 30% since he became mayor and people are coming back to San Francisco, leading to rising complaints about traffic and real estate values. He said he is working on stopping drug dealing on the streets and arresting people for illegal drug use. “Public transit, public safety and behavioral health,” are improving, he added.
He reluctantly answered a question about his role in keeping federal border agents from being sent to San Francisco.
Without mentioning him by name, Lurie pointed out that Salesforce CEO Mark Benioff, who The New York Times reported had concerns about security at the company’s Dreamforce conference that prompted his request for the National Guard to come to San Francisco, does not live in San Francisco.
Lurie said he offered to take Benioff with him as he walks through the streets meeting with people.
The final speaker was Ba, the xAI co-founder and Toronto professor who studied with 2024 Physics Nobel Prize laureate Jeffrey Hinton.
Newcomer asked Ba about xAI calling itself MechaHitler and whether Ba thinks Trump won the 2020 election. Ba’s responses, which seemed to view these incidents as opportunities to train xAI’s “truth seeking” models, struck me as non-responsive to Newcomer’s questions.
When asked what Elon Musk told him recently, Ba said “build human emulators.” I was thinking in the wake of his potential $1 trillion Tesla pay package, perhaps Musk is hoping Grok can someday control the army of robots he aspires to build.
I was also struck by Konwinski’s comments about how government cuts to research budgets are making Beijing the place where the world’s most talented researchers want to work because they can publish their work, build open source models, and “nudge the fate of humanity.”
Finally, I found it interesting that many of the speakers were talking about their children, how they were using AI and what kind of a future they were creating for them.
One of the speakers mentioned Musk’s comment about how smart people should have more children. This made me realize that children and the cost of living generally is so high, especially in San Francisco that only the wealthiest people can afford to own a home and raise children.
The future of generative AI could all depend on whether OpenAI CEO Sam Altman ends up being more like WeWork CEO Adam Neumann, who magically spun a money-losing business model to investors like Masayoshi Son, or Apple’s late CEO, Steve Jobs.
If Altman is more like Jobs, OpenAI could one day have a $1 trillion IPO. If Altman is more like Neumann, a less lucrative outcome could ensue.
