Will generative AI accelerate or slow down in 2025? Google CEO Sundar Pichai sees a slowdown while OpenAI CEO Sam Altman envisions an acceleration next year, according to the New York Times.
While this debate is worth examining, the more important question is: Will generative AI live up to its hype or fall short over the long run?
To that question, my answer is: It is too early to know. My book Brain Rush compares generative AI to previous hype-rich technologies such as the World Wide Web, blockchain, and virtual reality. Of these three, the World Wide Web came closest to living up to its hype.
Will generative AI ultimately fall way short of its hype — as seen with virtual reality, which has cost Meta Platforms $45 billion in losses so far — or achieve bullish prognostications as a growth and productivity booster?
To answer that question, leaders should seek answers to questions I will pose to Babson College students next month in my new course, Generative AI For Business Growth And Strategy.
Here are five questions leaders should answer — tailored to their specific organizations:
- Does the technology target an important unmet societal need?
- How significant is the technology’s profit potential?
- Conversely, how big a threat does the technology pose?
- Can the technology deliver a leap in value to customers compared to existing solutions?
- Does the company have — or can it obtain — the skills needed to prevail over rivals in providing that superior solution?
The answers to these questions are very different depending on whether a company is supplying the picks and shovels for generative AI or the company is mining for generative AI gold.
As for the suppliers, one company — Nvidia — stands above the rest in answering these questions in the affirmative. A few — such as Vertiv, ServiceNow, and Salesforce — are making good early progress as many suppliers still struggle.
Few companies are getting measurable value from AI chatbots. One that stands out is the Stockholm-based fintech Klarna, which is using generative AI to perform customer service that enables the company to operate with 22% fewer employees, according to Bloomberg.
Sadly, no company has ever cut its way to prosperity. For that, companies must use generative AI to create new growth curves — something few companies have yet accomplished, according to my Value Pyramid note.
What Google And OpenAI Expect For Generative AI In 2025
While Google sees a near-term slowdown in generative AI, OpenAI expects huge breakthroughs in 2025. Before getting into the reasons for these diverging forecasts, let’s look at where each company is positioned for generative AI growth potential.
Pichai’s company has shed some AI talent. The researchers who invented the Transformer models used to train and operate large language models have almost all left Google — reflecting poorly on the company’s ability to retain talent, Brain Rush noted. Still Google has vast resources including YouTube and Gmail, which provide data for LLM building, and a leading cloud service in Google Cloud.
Altman’s OpenAI has different strengths and weaknesses. The LLM developer depends on capital and cloud services from Microsoft — with which it has a complicated relationship — and projects losses of $44 billion through 2028, according to Foundation Capital. On the good side, Open AI has 300 million users and a $157 billion private market valuation on its $11.6 billion in projected 2025 sales.
Google offers models that compete with ChatGPT but Pichai does not expect big progress in 2025. “The hill is steeper … You’re definitely going to need deeper breakthroughs as we get to the next stage,” Pichai said at the December 4 DealBook Summit, noted the Times.
Google envisions the possibility of AI chatbots getting better at reasoning and agentic AI — “completing a sequence of actions more reliably,” Pichai said. He envisions such improvements offering corporate users a payoff from generative AI — despite investment expected to top $1 trillion, according to a Goldman Sachs report.
One reason for Pichai’s tempered 2025 outlook is OpenAI’s upcoming LLM, Orion, which “exhibited diminishing returns in performance compared to earlier models,” reported Dataconomy.
Altman disagrees with Pichai. “There is no wall,” Altman said, according to the Times, which noted that Microsoft’s exclusive license to his company’s technology ends when OpenAI says it has achieved AGI. While Altman says AGI will happen in 2025, his claims match what OpenAI researchers are saying “on the ground,” according to ReadWrite.
Generative AI Picks And Shovels: Hardware And Some Application Suppliers Making Bank
Nvidia shareholders have enjoyed a spectacular return on investment. Since May 2023 when Nvidia’s boffo growth forecast prompted me to write Brain Rush, the company’s stock has risen 244%.
Nvidia has gotten the answers right for those five questions I mentioned above. Cloud services providers have a huge unmet need for chips to train and operating LLMs. With 90% market share, Nvidia clearly has captured the opportunity by offering customers a significant return on the high investment in its chips, CEO Jensen Huang argued, according to my September 2024 Forbes post.
Since the company designed the chips and wrote the software developers use to train and operate LLMs, Nvidia has clearly assembled the strongest skill set in the industry for satisfying the needs of those cloud services providers.
Generative AI Miners: Very Few Companies Boosting Productivity
In the long run, demand for Nvidia chips will last as long as generative AI miners — meaning the companies trying to use generative AI to improve their operations — convert that improvement into more value.
Most companies have yet to capture such value. Indeed, no killer app — such as the electronic spreadsheet, which gave people a reason to buy personal computers back in the 1980s, or the iTunes store that boosted sales of the iPod in 2003 — has emerged for generative AI.
Many employees are using AI chatbots to overcome creator’s block — for example, a first draft of an email or a marketing pitch. A small number of companies are using the technology to make business functions — such as coding or customer service — more productive. Virtually no companies are using generative AI to create new growth curves to accelerate revenue, notes my Value Pyramid.
To be fair, Klarna has reduced headcount by 22% to 3,500 and now uses Gen AI to perform the functions of “hundreds of staff,” CEO Sebastian Siemiatkowski told Bloomberg. More specifically, Klarna’s OpenAI-powered AI assistant, which is doing the work of 700 full-time customer service agents.
Klarna — with 85 million customers and a valuation that plunged 68% from its 2021 peak to its recent $14.6 billion, noted Bloomberg — pledges to boost the salaries of the 200 employees using AI in their core work if they become more efficient, Siemiatkowski said.
Sadly, such cost savings will not boost Klarna’s value unless the fintech deploys generative AI to grow faster than rivals and exceed investor expectations.
Whether Klarna and most companies can deliver such results will determine whether history judges generative AI to be a costly fad — like virtual reality — or a transformational innovation like the World Wide Web.