Source: Wall Street News
By building huge infrastructure, tech giants deeply embed AI into every aspect of their products and services to create powerful new engines for profit growth. In contrast, startups are obsessed with developing extreme AI models and have fallen into the dilemma of “burning money” but not making money.
Over the past year and a half, investment in artificial intelligence has surged. Silicon Valley startups are obsessed with developing the ultimate AI model. The capital seemed to have been thrown into an incinerator, and instantly went up in smoke.
However, when investors' enthusiasm receded, people were shocked to discover that currently only giants that monopolize AI infrastructure (such as Nvidia, Google, and Microsoft) have made money, but downstream startups have fallen into the dilemma of “burning money” but not making money. Even the leading AI startups known as the “Four Little Dragons” — Anthropic, Stability AI, Inflection AI, and OpenAI — are struggling.
According to recent media reports, Inflection AI, which has a valuation of up to 1.5 billion US dollars, has almost no revenue. It has closed its original business, Stability AI has laid off employees and parted ways with its CEO. Anthropic is rushing to resolve the $1.8 billion annual revenue and expenditure gap, and OpenAI is facing challenges in terms of sales growth.
“The cruelty of reality is written on the wall,” said Ali Ghodsi, CEO of Databricks, a data warehouse and analytics company that partners with artificial intelligence startups. “No matter how cool something you've created, the question is whether it's commercially viable?”
The unicorns, whose valuations have soared, are still burning money!
According to PitchBook data, investors have injected capital into about 26,000 AI and machine learning startups in the past three years, reaching a scale of US$330 billion, which is two-thirds more than the amount they invested in 20,350 AI companies from 2018 to 2020.
The cost of developing AI is huge, and as investors' attitudes towards AI gradually become lukewarm, all startups are desperately looking for ways to cover current huge expenses and achieve profits in order to win back investors' confidence.
However, up to now, they have not come out of the crazy stage of burning money, and are still far from making a profit.
The media quoted sources as reporting that Anthropic has raised more than 7 billion US dollars with the support of Amazon and Google, and can burn 2 billion US dollars a year, while actual revenue is only 1.5-2 billion US dollars, and the annual income and expenditure gap is as high as 1.8 billion US dollars.
Anthropic is promoting partnerships with major tech companies to sell customized AI systems and chatbots to businesses and government agencies. It recently announced that it is partnering with global consulting firm Accenture to create custom chatbots and AI systems for businesses and government organizations.
Anthropic spokeswoman Sally Aldous said that currently thousands of companies are using the company's technology, and millions of consumers are using its chatbot Claude.
According to the report, Stability AI, which focuses on image generation, is expected to have sales of 60 million US dollars this year, while the cost of its image generation system is about 96 million US dollars. The system will go on sale in 2022.
Stability AI's financial position is better than other large model companies due to the lower cost of image generation systems, but payment requirements are uncertain. The founder of Stability AI had no choice but to resign due to operational pressure, and then the company embarked on a path of layoffs and restructuring.
Notably, Stability AI has been operating without the support of tech giants, raising $101 million from venture capitalists in 2022. The report said that Stability AI needed more capital last fall, but it is difficult to prove to investors that it can sell technology to companies. It raised $50 million from Intel late last year, but despite this, it is facing significant financial pressure.
Inflection AI, one of the most highly valued AI startups in Silicon Valley with a valuation of more than 1.5 billion US dollars, was also “dug up” by Microsoft due to lack of commercial returns, and the founders team was forced to return to the big factory system to survive.
An investor revealed to the media that Inflection AI generated almost no revenue a year after launching the AI Personal Assistant. This also means that unless it continues to raise huge sums of money, it can't continue to upgrade technology and compete with chatbots from companies like Google and OpenAI.
Inflection AI CEO Mustafa Suleyman rose to fame as one of the founders of DeepMind. Suleyman founded Inflection AI with Karén Simonyan, former DeepMind researcher, and Reid Hoffman, a well-known venture capital in Silicon Valley.
Microsoft used to be the financier of Inflection AI. In March, Microsoft announced that Suleyman and Simonyan would leave the company and go to Microsoft to form a new team called “Microsoft AI.” Most of Inflection AI's other employees have also been taken into Microsoft's pocket.
The report said it would cost Microsoft more than 650 million US dollars, but unlike Inflection AI, it can play a protracted “money-burning” game.
OpenAI gained popularity with ChatGPT, but like many other AI startups, it's also struggling to find a way to profit.
Two people familiar with the company's business revealed to the media that OpenAI made about $1.6 billion in revenue in the past year, but it's unclear how much the company spent.
Allegedly, OpenAI is facing challenges in expanding sales, and enterprise customers are concerned that GPT may output inaccurate answers. Additionally, OpenAI has been questioned over whether the data the model relies on infringes copyright.
Compared to an annual revenue of less than $2 billion, OpenAI's valuation has reached an astonishing $100 billion. Furthermore, according to the “2024 Global Unicorn List” released by Hurun Research Institute, OpenAI's valuation reached 710 billion yuan, making it the fastest growing company in value, with an increase of nearly 570 billion yuan.
Microsoft has invested $13 billion in OpenAI and owns 49% of the company's shares. In February, OpenAI allowed some employees to sell their shares at a valuation of $86 billion through a takeover offer.
How do tech giants make money from AI? Integrate AI into every product
Unlike startups that focus on developing cutting-edge AI “breakthrough” technology, tech giants use AI as a “paving stone”. By building huge infrastructure, AI is deeply embedded into every aspect of their products and services to create a powerful new engine for profit growth.
Giants such as Microsoft, Google, and Nvidia monopolize AI productivity factors, including computing power, chips, and data, to build barriers to AI technology and lay a solid foundation for their continued profits.
As one of the first tech giants to enter the generative AI market, Microsoft is the best example of AI monetization. Brad Reback, an analyst at investment bank Stifel, said that AI services in Microsoft cloud computing contributed about 1 billion US dollars in sales in the first quarter, compared to almost zero a year ago.
On the other hand, major giants are still increasing their investment in AI. During the first quarter earnings call, Meta, Google, and Microsoft all emphasized their AI investment expectations this year:
Meta raised its capital expenditure forecast for this year by 10 billion US dollars, rising to 35 billion to 40 billion US dollars for the whole year;
Google says it will spend about 12 billion US dollars or more on capital expenses every quarter this year;
Microsoft says its capital expenditure for the most recent quarter was 14 billion US dollars, and it is expected that it will continue to increase “significantly.”
Judging from the first-quarter results and forecasts already announced by major technology companies, a company must spend $10 billion every quarter to take a place in the AI field.
In the new era of big models, small companies stand on the sidelines?
Morgan Stanley previously pointed out that the world is entering a new era of rapid growth in the capabilities of big models driven by hardware and software, and that big models will significantly improve their creativity, strategic thinking, and ability to handle complex multi-dimensional tasks.
However, the high cost of the supercomputers needed to train the next generation of big models is a huge challenge even for tech giants, let alone small companies.
In addition to high capital expenses, barriers to chip power supply and artificial intelligence technology are also increasing. Together, these factors form a major obstacle to entering the big model field, and may make it difficult for small companies to compete with powerful giants.
edit/lambor