Today, the field of artificial general intelligence (AGI) has announced a series of blockbuster news.
In the early morning of April 6, Beijing time, according to the Wall Street Journal, xAI, an AI startup founded by billionaire and Tesla CEO Elon Musk, is planning to raise $3 billion, and this round of financing will bring the company's valuation to $18 billion.
The report quoted investors close to Musk as saying that the financing partners in this round include the Gigafund fund and venture capitalist Steve Jurvetson. However, the terms of xAI's financing have not yet been finalized, and plans are subject to change.
Meanwhile, according to The Information, OpenAI co-founder and CEO Sam Altman and Apple's former chief designer Jony Ive plan to co-launch a new AI hardware company and are seeking $1 billion in funding, with potential investors including OpenAI's existing investor Thrive Capital, SoftBank CEO Masayoshi Son, and Jobs' widow Lauren Powell. Emerson Collective Fund, founded by Steve Jobs, etc.
AI in the United States continues to "soar", and Musk and Altman have raised a total of more than 4 billion US dollars (about 21.699 billion yuan), so as to create a new world of AGI faced by mankind in the future.
"Jony Iveive"s iPhone Ive 100 y ai ai y 1 y 1 y 1 y 1 y y 1 y y 1 y.
Is the "AI iPhone" really coming?
On April 6, it was reported that Altman, a mysterious AI hardware company co-founded with former Apple design director Jony Ive, is currently seeking outside investment to raise up to $1 billion.
According to the report, the company has held discussions with Emerson Collective, led by Jobs' widow Laurene Powell Jobs, and Thrive Capital, a significant investor in OpenAI. In addition, the two are in talks with SoftBank CEO Masayoshi Son to fund their upcoming AI equipment business.
But it's unclear if OpenAI will own part of the Ive-Altman joint venture. Spokespeople for Thrive and Emerson did not comment. SoftBank CEO Masayoshi Son has been in talks with Altman and Jony Ive since the company's inception, but his involvement cannot be known.
It is reported that Jony Ive is regarded by the outside world as one of the core figures of Apple's products, mainly promoting the development of iPhone design, and is known as the "father of iPhone". He left Apple in 2019 to start his own design studio, LoveFrom.
As early as last year, the Financial Times and other sources said that Altman and Jony Ive hope to create an AI device, and the AI device to be developed is not expected to look similar to a traditional mobile phone, according to previously disclosed information, the new product will be ARM's chip, Apple's hardware, OpenAI's software triad.
But some investors have warned that it may be difficult to support the company for now. Considering the addition of Jony Ive and Altman, the company could be valued at a higher starting price. However, the company is still in its infancy, in a field that has not yet been proven, and there are no physical products.
At the investor level, The information discloses the relationships of three institutions:
- Lauren Jobs had a close personal relationship with Altman and Jony Ive, whom she had known for decades. Emerson Collective is a philanthropic organization and venture capital firm that has invested in online payments company Stripe, financial technology company Chime, and HR software seller Gusto, among others.
- And Thrive recently spearheaded the purchase of up to $1 billion worth of OpenAI shares, including shares from employees, a price that values the company at $86 billion. A year ago, Thrive bought a large amount of OpenAI stock for about $27 billion. Thrive is also a majority shareholder in Stripe and credit card startup Ramp.
- Son hopes that chip design company Arm will be involved and play a big role in this AI device project.
In fact, tech companies are stepping up the development of new hardware products that integrate with AI to bring AI to life in the consumer space.
Altman is also one of Humane's largest investors, including a pin-style wearable device with a camera, microphone and speaker, and a device called Rabbit, which helps users access services across apps.
Meanwhile, OpenAI recently discussed embedding its object recognition software into Snapchat's parent company's products, potentially bringing new features to Snap's Spectacles smart glasses.
In addition, large language model developers such as OpenAI, Google, and Meta Platforms are also working to improve the capabilities of their models for image, drawing, and gesture recognition, rather than just language recognition. The developers hope that the integration of the new technology will enable large voice models to be configured into smart glasses and other wearables with front-facing cameras.
Market analysts believe that the ultimate goal of all efforts may be aimed at creating AI assistants that have only appeared in science fiction movies before. They can serve as tutoring for students when writing essays or solving math problems, or to help users when they ask for information about their surroundings.
According to multiple market sources, Altman is planning to make a large investment in the field of AI computing power, including $7 trillion to build an AI supercomputing network, $115 billion to build a "stargate" with Microsoft, and to launch a less powerful data center in Wisconsin as soon as 2026.
Musk's "three steps": raise $3 billion, raise salaries to recruit AI talents, and launch Robotaxi in August
Seeing that OpenAI raised $13 billion last year and was valued at more than $80 billion, but the company's founder Musk failed to get benefits, Musk is really anxious now.
Musk (source: video screenshot)
According to the Wall Street Journal, investors close to Musk are in talks to help raise $3 billion for his AI startup xAI, which will value the company at $18 billion, according to people familiar with the matter. It is reported that the backers who are considering investing in this round of financing include venture capital firm Gigafund and American venture capitalist Steve Jurvetson.
According to the Wall Street Journal, the proposed fundraising marks another escalation of the AI arms race. In recent years, leading AI startups have raised tens of billions of dollars, such as OpenAI raising $13 billion and Anthropic raising more than $6 billion, to train AI large model technology and generative AI products.
It is reported that in July 2023, Musk announced the establishment of a new AI company, xAI, with a mission to "understand the true nature of the universe", with the goal of building a competitor to OpenAI. Last fall, Musk said that X Corp. investors would own a 25% stake in xAI.
As one of the founders of OpenAI and the latest critic of OpenAI, Musk has previously said that OpenAI has deviated from its intended purpose and become a profit-oriented entity. He bluntly stated that OpenAI was originally created as a non-profit open-source organization to compete with Google. But it has since become a closed-source, profit-oriented company under Microsoft's control. "OpenAI has become a profit-maximizing company...... This was not his original intention at all. ”
Musk also denounced OpenAI being controlled by Microsoft. He said the world needs an alternative to GPT's AI technology options.
In November last year, Musk officially released xAI's first large model and application result solution, Grok, and built the Grok AI assistant on the social platform X, Musk has said that in some ways, it is the best (AI technology) that currently exists. The tuned Grok model has 33 billion parameters, which is close to Llama 2 -70B in terms of performance benchmark, and the results are 63.2% and 73% on the HumanEval encoding task and MMLU benchmark, respectively.
In March, xAI unveiled its latest Grok-1.5 model and said it would soon be available to early testers as well as existing Grok users.
According to Musk's filings with the Securities and Exchange Commission (SEC), in December, xAI was looking to raise $1 billion and has already raised nearly $135 million from four investors. It is not yet possible to determine whether the fundraising is part of the ongoing negotiations.
At the time, Musk denied previous reports that he was raising additional money for xAI. "xAI hasn't raised money, and I've had conversations with anyone about it," he posted on X in late January.
The report quoted people familiar with the matter as saying that the current financing negotiations for xAI are gaining momentum, and not only Feng County Investment Fund is needed, but more special investment funds are needed. Among them, Steve Jurvetson is an old friend of Musk, a member of the board of directors of SpaceX, and a former director of Tesla; Gigafund was co-founded by Luke Nosek, who is also a director of SpaceX and founded PayPal with Musk, with whom Musk also tried to buy DeepMind, which is now part of Google.
But for this financing, as of now, Musk and xAI and X platform have no comment.
At the same time, Musk has complained that the company is raising the pay of AI engineers to prevent companies like OpenAI from poaching.
He said Tesla is raising salaries, while OpenAI is "aggressively recruiting Tesla engineers with huge pay." He believes that the competition for AI engineers is "the craziest battle for talent I've ever seen."
According to reports, Tesla machine learning scientist Ethan Knight left Tesla to join Musk's AI startup xAI. Musk responded: "Ethan was going to join OpenAI, so it's either xAI or them." ”
OpenAI did not immediately respond to a request for comment.
Recently, the boom in AI has intensified the war for talent in Silicon Valley, with startups competing with established tech giants to build the next "big thing." Tech companies are rolling out million-dollar annual salary packages, accelerated stock vesting plans, and poaching entire engineering teams in an attempt to attract talent with expertise and experience in generative AI to fuel their AI products.
Musk has said that don't think of Tesla as a car company, it's more of an AI robotics company.
Currently, Tesla's work on AI is mainly focused on driverless cars and humanoid robots. In 2023, Musk revealed that Tesla plans to spend more than $1 billion to buy a supercomputer, named Dojo, by the end of 2024 to develop driverless technology.
In addition to raising $3 billion and raising salaries to recruit AI talents, on April 5, Musk also announced a bombshell news: Tesla will release Robotaxi - driverless taxis on August 8, but he did not disclose more specific information.
If Robotaxi can be released as scheduled, it will be tantamount to an industry earthquake. There is a consensus in the automotive industry that autonomous driving technology will disrupt existing business models and disrupt the industry landscape.
Musk has always wanted to scale up EV sales with cheaper product platforms. As early as 2020, he proposed that Tesla's goal is to maintain an average annual sales growth rate of 50% and reach 20 million units by 2030. Toyota is the world's No. 1 car company in terms of sales, with sales of 11.23 million units in 2023. Tesla's sales target is equivalent to double Toyota's current sales.
The next-generation product platform is crucial to Tesla, and there has been a debate within the company about it. Musk's biography, published in September 2023, describes the process of adjusting the new platform.
When the new product platform project was established in November 2021, Musk demanded that the steering wheel and brake pedal be eliminated, and the product should fully bet on autonomous driving. In Tesla's announcement, the earliest description of the next-generation product platform is also Robotaxi.
Tesla's team of engineers is concerned about the progress of self-driving technology and regulatory risks, and in August 2022 suggested that Musk should not be too aggressive and that the steering wheel and brake pedal should be retained for the new product.
According to the biography, Musk was not receptive to the suggestion at first, telling the team that Robotaxi would make other cars unnecessary. But Tesla designer Franz von Holzhausen did not give up convincing Musk, and he quietly kept and designed a new model with a steering wheel and brake pedal.
Just two days ago, Reuters reported that Tesla canceled the development plan for the next generation model with a target price of $25,000 and instead developed Robotaxi based on the platform. According to Reuters, Tesla's cancellation of the entry-level model program may have been influenced by Chinese competitors, and many products from companies such as BYD flooded the segment. Musk denied the above analysis, but then sent out a Robotaxi message.
Now, the Robotaxi project may be implemented. However, whether it can be carried out as scheduled depends on the maturity of Tesla's intelligent driving system FSD (Full-Self Driving). FSD has made significant progress recently. On March 31, Tesla pushed the new version of FSD V12 to some users in the United States. For the first time, the word "Beta" has been removed and replaced with "Supervised", which means "supervised".
The latest version of FSD uses "end-to-end" technology. Early beta users generally praised the new version as "more like a human driving a car". "End-to-end" means that the intelligent driving system inputs information from the perception end and directly outputs the results at the execution end. Its training method is similar to that of ChatGPT, a large natural language model.
Musk is extremely enthusiastic about self-driving technology, and he has repeatedly predicted the timing of the technology's mass adoption. Although previous predictions have failed, Musk is not reconciled, and he always throws out new predictions again.
Musk insists that the FSD system will outperform human drivers. He also constantly emphasized that Tesla's value lies in autonomous driving, otherwise it would be worthless.
The U.S. AI race has escalated, with 2,011 AI companies around the world continuing to compete
In addition to the above news, there are more AGI field news coming out today, such as Microsoft, AI data, AI technology trends, etc.
First, on April 6, Microsoft and Quantinuum jointly announced that their researchers would reduce the "noise" that causes huge errors in quantum computing by a factor of 800. This breakthrough resulted in more than 14,000 individual experiments performed without a single error by applying a qubit virtualization system and error diagnosis and correction techniques to Quantinuum's "ion trap" hardware. "This achievement is a critical step towards building a hybrid supercomputing system that can transform research and innovation across multiple industries. Microsoft said.
Second, tech giants are secretly competing for the secret track of more than $30 billion in AI training data.
Reuters reported that Ted Leonard, CEO of Photobucket, a traditional U.S. image hosting site with 70 million users, revealed that the company is in discussions with several tech giants to plan hundreds of millions of dollars to license Photobucket's 13 billion photos and videos to model companies to train generative AI models that can create new content based on text prompts.
You can understand that the American version of Visual China wants to sell all the images on the website to large model companies to train and generate AI image or video content that everyone can imagine.
Price-wise, with prices ranging from 5 cents to $1 per photo and over $1 per video, prices vary depending on the buyer and the type of image being sought.
Jarrod Yahes, Shutterstock's chief financial officer, told Reuters that its big tech giants initially amounted to $25 million to $50 million per deal, but most have since increased. Smaller tech companies have followed suit, sparking a new "flurry of activity" over the past two months, he added.
Seattle-based Defined.ai CEO Daniela Braga said the company licenses AI data to a range of companies, including Google, Meta, Apple, Amazon and Microsoft. Rates vary depending on the buyer and the type of content, but companies are typically willing to pay $1 to $2 per image, $2 to $4 per clip, and $100 to $300 per longer film. She added that the price of text data is $0.001 per word. Whereas, nude images that require the most sensitive processing are sold for $5 to $7.
Many major market research firms say they struggle to estimate the size of the AI data market because many companies are unable to disclose specific information. Researchers such as Business Research Insights estimate the market to be around $2.5 billion currently and predict that it could grow to $30 billion within a decade.
Edward Klaris of Klaris Law said: "There is now a lot of interest in finding copyright holders because they have private collections of material that cannot be removed. The law firm says it is consulting with content owners worth tens of millions of dollars, and everyone can license AI training for photos, movies and book archives.
Again, you know, how many of the world's leading companies in the field of AI are? Recently, there is a relatively accurate data: 2011.
A few days ago, Matt Turck, an investor in FirstMark, an American venture capital institution, jointly released a new article on MAD (Machine Learning, Artificial Intelligence and Data).
Matt Turck disclosed that as of March 2024, there are 2,011 AI-related companies in the world. Compared to 1,416 last year, 578 new companies were added, while only 139 companies were shown in the 2012 report.
According to the report, there have been two waves of AI in the past 60 years – a data infrastructure cycle of about 10 years, and an ML/AI cycle, and now two big waves and limited integration, with a large number of AI companies. The explosion of AI in 2023 is driving the biggest structural change to date in the world.
Phenomenally, last year, every technology vendor scrambled to integrate generative AI into their products. At the same time, the boards of directors of Fortune 2000 companies have instructed their teams to actively "apply AI" such as Morgan Stanley and Citibank, and have even deployed the technology in record time. In addition, consumer enthusiasm for generative AI applications is at an all-time high.
At the financing level, the current financing environment presents a "split in two" situation, and the AI field is completely different from other fields. Overall, global funding in AI fell 42% to $248.4 billion in 2023. In the early months of 2024, while there were several positive signs, the overall trend did not change significantly.
Specifically, several distinguishing features have emerged in the AI financing market, including:
- 巨额资本主要流向少数几家初创公司,尤其是 OpenAI、Anthropic、Inflection AI、Mistral 等。
- The participation of corporate investors is unusually high. The three companies that have invested the most in AI in 2023 are Microsoft, Google, and NVIDIA.
- In the above-mentioned corporate transactions, there is some uncertainty about the ratio of actual cash to "computing power in exchange for equity".
Overall, 2023 is a breakthrough year for companies in the AI space: OpenAI has reached $2 billion in annual revenue, Anthropic expects $850 million in revenue in 2024, Midjourney has soared to $200 million in revenue with zero investment and a team of only 40 people, and Perplexity AI has soared from zero to 10 million monthly active users, among others.
"We are living in a particular time, at the beginning of a paradigm shift. There's never been a better time to try something new. Our journey has just begun. The report ended with a statement.
(This article was first published on the Titanium Media App, author: Lin Zhijia)