the author is the founder Siftedan FT-backed site about European startups
The recent California Gold Rush paints a mad scramble for generative AI. US technology giants such as Google, Microsoft, Meta, Palantir, and numerous venture capital firms are all eagerly seeking new layers of digital treasure. But the big, and still unanswered, question is who will eventually spit the dust, and who will bag the most gold.
The obvious guess is that the big companies developing these text, image, video and audio generation models will dominate the field. As Federal Trade Commission Chairman Rina Khan writes, a few powerful companies control all the necessary raw materials: vast amounts of data storage, computing power and cloud services. It seems that. She may have added: They have many of the world’s leading AI researchers and have a lot of money.
“The growing adoption of AI risks further entrenching the market dominance of incumbent tech giants,” Khan recently wrote in The New York Times. Such a worldview further drives Khan’s willingness to break trust.
But judging by a leaked memo entitled “We Have No Moat” by one of Google’s executives, the world doesn’t look that way to some insiders of big tech companies. The executive noted in April that Google and OpenAI, with significant Microsoft support, may have developed some of the most capable closed generative AI models such as Bard and GPT-4. But they were already in danger of being overtaken by more agile competitors who are building smaller, cheaper, customizable open source AI models and poaching some of Google’s brightest researchers. rice field. “The uncomfortable truth is that we are not in a position to win this arms race, and neither is OpenAI,” the executive wrote. “Who would pay for a limited-use Google product when there are free, high-quality alternatives?”
This week, Google announced it would be building generative AI into its wider range of services, in an attempt to gain a competitive edge in its quest to catch up to Microsoft. But according to Google’s memo, the main beneficiary of the trend toward an open-source model could be Meta, which is also pivoting to AI. Meta, which launched its own open-source LLaMA model in February, now aims to build a platform for others to play on. Much like Google built a new app ecosystem around its open-source Android phone software, Meta could emerge as a platform for innovation to happen. “The only clear winner in all of this is Meta,” wrote a Google executive.
Meanwhile, venture capital investors are betting that a new wave of generative AI startups such as Anthropic, Cohere, Stability AI, Inflection and AI21 Labs will also find gold. Their logic is that at least some of these start-ups can move faster than the big players, monopolize some niche markets, and largely ignore costly safety controls (which should alert regulators). is.
Yoav Shoham, co-founder of Israeli startup AI21 Labs, said: “The future is for smaller, specialized generative AI that is cheaper to train, faster to run, and capable of serving specific use cases. I will be a model,” he said. “The moat isn’t the technology. It’s the relationship with the consumer,” Shoham tells me. “I think it’s going to be a ‘few wins’ market, not a ‘winner takes all’ market.” Many other start-ups that only offer generic services and don’t have customer buy-in will fail.
Incumbents in most industries that can feed their own data into the generative AI model and fine-tune the output could also grow. Their challenge is to redesign their organizational structure to take advantage of new technologies. Other clear winners are the “pick and shovel” companies that provide the tools for this technological transformation. The big cloud computing companies AWS, Google, and Microsoft will benefit from the voracious demand for computing power in this model. But his Nvidia, the leading designer of the graphics processing units that power most AI models, also stands out. “We are in an AI moment for the iPhone,” said Nvidia CEO Jensen Huang.
However, the pace of evolution in this field is so rapid that today’s best guess could turn into a piecemeal betting vote tomorrow. The history of other general-purpose technologies such as electricity, automobiles, and the Internet suggests that generative AI will create new markets and business models that no one can imagine today. Companies that have not yet been founded may be mining the most gold.
