Investors start asking hard questions about AI; Alexa's new brain is stuck in the lab; Arm may be the unexpected winner of the AI boom; meet the AI "super users"

Middle East wants to become the next global AI hub; what teens think of AI; AI automation comes for salespeople and recruiters; Meta takes on Perplexity; Visa finds hundreds of use cases for AI;

Nov 01, 2024

∙ Paid

It was earnings week for the Magnificent 7 and, despite record-breaking profits, investors had just one question on their minds: what is $200 billion of investment going to deliver in revenue when the next generation of AI arrives?

When OpenAI launched ChatGPT, everyone wanted to see what the hundreds of millions of dollars that went into training GPT-3 and GPT-4 could achieve. For OpenAI, the payoff has been substantial: today, the company’s revenue has peaked at $5 billion, proving that large-scale investment in AI infrastructure can drive solid returns, especially with a well-strategized product suite and a growing user base.

Now, the stakes are higher than ever. Amazon, Google, Meta and Microsoft have all indicated this week that they ramping up spending beyond the hundreds of billions they poured into AI this year, hinting at an industry-wide push towards the next era of foundation models. But this time, the anticipated price tag is nearly a trillion dollars, dwarfing previous expenditures and raising a critical question: where will the revenue for these next-generation models come from?

The scale of investment we’re witnessing now isn’t about incremental improvements; it’s about transformative leaps. These foundation models could shape entire industries, bringing us closer to AI that can interact, reason, and solve problems almost as intuitively as humans do. But scaling these capabilities will require infrastructure that’s orders of magnitude larger and more sophisticated than what we have today. Training next generation models demands not only vast computational resources but also access to high-quality, proprietary datasets—an expensive, time-intensive endeavor that only a handful of companies have the resources to undertake.

So the logic goes that if Big Tech companies are willing to collectively front a trillion dollars in capex, it’s because they foresee a payoff that matches or beats the 10x investment-to-revenue ratio achieved by OpenAI in 2023-2024; in short, that we’ll move from hundreds of millions in, billions out to trillions in, tens of trillions out.

One potential revenue stream is enterprise licensing. Just as OpenAI charges businesses to integrate GPT-4 into their products, these companies could sell exclusive access to AI tools that drastically streamline operations. However, given the upfront costs, this traditional licensing model may not deliver sufficient returns to justify a trillion in expenses.

The future may lie in scaling down to scale up. Many companies might opt for a decentralized approach, democratizing access to their models while allowing developers to customize and extend the core functionalities. This could unleash a wave of applications—from industry-specific AI to personalized assistants—that would rely on API access. For tech companies, this could mean sustained subscription revenue, an approach that has proven effective for GPT-4, as well as royalties and licensing fees from a proliferation of microservices.

While there’s no definitive playbook for recouping such a monumental investment (the US spends about $800 billion a year for defense, for reference), the willingness to gamble on AI suggests that Big Tech sees this as a once-in-a-century opportunity to redefine their industries—and finally make technology the industry to rule them all. However, the bet isn’t without risks. It’s a financial and logistical arms race, and any misstep could result in colossal losses rather than the anticipated leap forward.

But there’s another question that investors (and especially VCs) should also probably ask themselves: what will happen to the dozens of LLM startups that received hundreds of millions in funding and burnt through most of that this year to build almost-as-good models to today’s Llama, GPT-4 or Gemini while achieving only millions of dollars in revenue? Will they also get a trillion dollars and a second shot to compete with the Big Tech AI labs? And how will they monetize their research this time around?

And now, here are this week’s news:

❤️Computer loves

Our top news picks for the week - your essential reading from the world of AI

FT: Wall Street frets over Big Tech’s $200bn AI spending splurge
Wired: This Is a Glimpse of the Future of AI Robots
Bloomberg: Alexa’s New AI Brain Is Stuck in the Lab
FT: How Arm could be the unexpected winner of the AI investment boom
MIT Technology Review: Cultivating the next generation of AI innovators in a global tech hub
Reuters: Saudi’s Davos is no longer such a desert
Wired: AI Will Understand Humans Better Than Humans Do
Time: What Teenagers Really Think About AI
The Information: Seven Acquisitions That Could Happen in 2025
Washington Post: Meet the ‘super users’ who tap AI to get ahead at work
The Information: ‘Fear and Skepticism’: AI Automation Arrives for Salespeople
MIT Technology Review: Palmer Luckey on the Pentagon’s future of mixed reality
The Information: Meta Develops AI Search Engine to Lessen Reliance on Google, Microsoft
TechCrunch: We finally have an ‘official’ definition for open source AI
The Verge: Google plans to announce its next Gemini model soon
Forbes: He Sold His AI Company To AMD For $665 Million. Now He Wants To Supercharge AI Research In Europe

⚙️Computer does

AI in the wild: how artificial intelligence is used across industry, from the internet, social media, and retail to transportation, healthcare, banking, and more

🧑‍🎓Computer learns

Interesting trends and developments from various AI fields, companies and people

Keep reading with a 7-day free trial

Subscribe to Computerspeak by Alexandru Voica to keep reading this post and get 7 days of free access to the full post archives.