The initial surge of the AI boom ignited fierce competition between new startups and established industry giants. However, this rivalry has recently cooled, evolving into a tug-of-war among a few leading players: Google, Anthropic, and OpenAI hold the top positions, followed by others like Meta and SpaceX's AI division.
Many believe the vast majority of the industry's spoils will ultimately be captured by these giants. Veteran tech investor and industry executive Josh Elman holds a starkly contrasting view. He is convinced a wave of explosive growth for consumer-facing AI startups is imminent. This conviction led him to leave Apple's AI transformation project several weeks ago to join Andreessen Horowitz (a16z), focusing exclusively on consumer sector investments.
Elman is acutely aware of the industry landscape he is entering. Before Apple, he spent years at venture capital firm Greylock, leading investments in prominent projects like Discord, Musical.ly (the precursor to TikTok), and Meerkat (later renamed Houseparty and acquired by Epic Games). Prior to Greylock, he held positions at LinkedIn, Facebook, and Twitter.
To counter a pessimistic outlook, Elman draws a parallel between the current state of the AI industry and the early days of the dot-com bubble.
"In terms of mass adoption, today's AI industry is at most at the level of the internet in 1995, or even 1996," he says. What follows is a condensed and polished transcript of the interview.
Interview Transcript
Interviewer: Let's start with the question on everyone's mind: Is Siri actually good now?
Elman: (Laughs) The public's core demand from Siri has always been simple: understand user commands and smoothly help complete various tasks on the phone. At this stage, Siri does that quite well. But if you compare Siri to a Claude office assistant, asking it to generate a complex presentation on a specialized topic, that's not Siri's current design purpose.
However, it accurately responds to daily commands. For example, if I say, "Call Abe who texted me this morning," the system dials directly. I always forget people's last names, so I use this feature daily. Another example: I was driving to Palo Alto for dinner. After exiting the highway, I had two route options. I simply gave the voice command, "Navigate to the restaurant where I'm meeting John for dinner," and the system immediately plotted the optimal route directly to the location. I hadn't even saved the restaurant's address in my calendar; we had only confirmed the plan via text.
Interviewer: Apple's AI business is currently thriving. Why leave now to return to venture capital, specifically focusing on consumer startups?
When I left Greylock in 2020, I believed the consumer internet sector's growth potential had bottomed out, making it difficult to find high-quality new projects. While a few products emerged during that time, like the live-streaming commerce platform Whatnot (invested in by a16z) and the AI music generator Suno, high-quality consumer AI products were overall scarce. The prevailing industry consensus was that core AI innovation for the next decade would be firmly controlled by foundational model companies like OpenAI, Anthropic, and Google.
Interviewer: What makes you so certain the mainstream industry view is wrong?
About six or seven months ago, when OpenClaw launched, it clicked for me. Agentic AI has the potential to reshape a massive number of consumer products. The consumer domain is poised for a Cambrian explosion of technology, and this wave hasn't truly arrived yet. Right now is the golden window for positioning.
Interviewer: But OpenClaw's current popularity is largely concentrated within the Silicon Valley professional circle, with most people using it to create office tools.
I've seen many excellent real-world use cases of ordinary users employing OpenClaw. For instance, two mothers, Claire Vo and Jesse Genet, use this tool to manage their family's daily affairs and care for their children. (Vo is the founder of ChatPRD; Genet is the co-founder and CEO of home goods brand Lumi).
I predict that over the next six months, a batch of promising prototype consumer AI products will emerge. It's still unclear which one will break out first. But within two years, various dedicated agent products for personal life management will be launched in bulk. Users will also be able to fine-tune these models themselves to match their personalized needs.
Interviewer: Consumer agents are certainly appealing, but won't the giants with massive user bases and solid brand foundations—Anthropic, OpenAI, Google, and even Apple—monopolize this track?
Whether it's finance, travel, or health management, people currently do use ChatGPT or Google searches to self-diagnose health issues. But in the future, products with far superior user experience and richer functionality will emerge, and these new products won't come solely from the major players.
For example, there are at least five mainstream code assistant tools on the market now, making that segment highly competitive. Yet, the travel sector still lacks a mature, user-friendly, all-in-one AI product; that market is essentially a blank slate.
My time at Apple gave me a profound understanding: No matter how large a company is or how ambitious its goals, it is remarkably difficult for one firm to execute a limited number of things perfectly. Companies prefer to deepen their core business and refine their flagship products rather than trying to cover every possible sector. There is a vast gap between the capabilities of giants and the full spectrum of market needs. This gap represents a multitude of opportunities for startups.
Comments