Weekly Series #5
Topics: Elon Musk's xAI raises $6B, OpenAI in news again, Salesforce loses $40B in valuation, Google AI Overview shows odd results
Elon Musk’s AI startup ‘xAI‘ raises $6B:
xAI announced raising $6B in series B funding from big name investors including Sequioa Capital, Andreessen Horowitz, Saudi Arabian Prince Al Waleed bin Talal among others.
This funding round brings xAI’s valuation at a whopping $24B, making it one of the most valued AI startups in the world. This fund will be used to propel AI research and build infrastructure that includes thousands of NVIDIA H100 GPUs in what might be the biggest cluster ever created, dubbed as “gigafactory of compute” by Elon Musk.
xAI was founded in July 2023 with a mission of “understanding true nature of the universe“. Their latest product Grok 1.5, a conversational chatbot similar to the likes of ChatGPT, is available to all X (formerly Twitter) premium users.
Salesforce sees biggest drop in stock price in over 2 decades:
Salesforce held Q1 earnings call on Wednesday last week and announced Q1 revenue of $9.13B, which happened to be 0.3% below analysts’ expectations.
The market had a rather strong reaction for a fractional miss in earnings expectations, pulling down Salesforce stock price by 19.7% on the following day, it’s biggest drop since 2004.
Moreover, Salesforce outlook for Q2 was lower than analysts’ expectations and since stocks are valued based on future expectations, the stock took a massive hit. Although, Salesforce still maintained its full-year forecast of $37.7-38B in revenue, which is 8-9% growth Year-over-Year.
Salesforce stated that they were seeing higher budget scrutiny from customers along with elongated deals and deal compression, leading to a softening outlook. However, their AI product Einstein AI is catching momentum according to their statement.
Salesforce soft outlook also dragged many other SaaS companies down along with it, which included Veeva, Snowflake, MongoDB, etc. as enterprises are expected to deploy capital cautiously on their product offerings.
OpenAI in the ‘news’ again:
The company continues to capture media attention on a variety of topics, soon after announcing their new model ChatGPT-4o and the scuffle with Scarlett Johannson that followed over voice resemblance. Here’s a breakdown:
OpenAI signed deals with The Atlantic and Vox Media on Wednesday last week, that will enable OpenAI to license their editorial content to train its models.
OpenAI already have signed deals with Reddit, NewsCorp, Axel Springer, DotDash Meredith, Financial Times and The Associated Press
The company has been quickly signing multi-million dollar deals with major content publications after being sued by New York Times for copyright infringement, along with others - Chicago Tribune, NY Daily News, Intercept.
Jan Lieke, OpenAI’s former co-lead of “Superalignment” team, joins rival company Anthropic AI. Jan and Ilya Sutskever, both announced their departure from OpenAI, citing disappointment in the organization’s direction to deprioritize AI safety research.
Google’s AI messes up:
Soon after a glamorous developer conference Google I/O few days ago, the company started rolling out its new search feature “AI Overview“ for all its users - a feature that used the content from top links and provided a short summary to the user upfront, removing the need for the user to visit those individual pages.
Soon after, users started reporting odd and inaccurate results coming out from this feature. Here’s an example where it recommended users to use glue to prevent cheese from sliding off of a pizza:
Many users started posting these anomalies online and it soon turned into a trend. Google published a blog last week acknowledging and explaining this odd behavior.
This was the second time this year where Google’s AI started showing cracks soon after general availability, the first one being early this year about showing a ‘woke’ bias with image generation.
