Way back in 2014, Meta/Facebook acquired WhatsApp for $19 billion, and just recently, they decided to monetize their 2 billion WhatsApp user community with ads. They also just plunked down $14b on Scale AI.
Even bigger, perhaps, is Apple thinking about buying Perplexity, which, to me, makes a lot of sense, especially because of their mishaps with Siri and “Gpt-ing” on the iPhone. Derek David of Medium concurs, saying, “Perplexity is doomed” in competing with Google and its leading Gemini model. However, a U.S. District Court just ruled in favor of the Department of Justice for Google’s alleged violation of antitrust laws by monopolizing open-web digital advertising markets. Since Google currently uses search revenue to fund things like Gemini and Google Cloud Services (GCS) to compete with AWS and Azure, if the Department of Justice prevails with their case, Google may have to split up, which would ignite a spending spree to compete.
Regardless, throwing antitrust behavior to the winds, Google is paying $2.4b to hire key staffers from startup Windsurf, to work on agentic coding at DeepMind and focus mostly on Gemini. Google won’t have any control or stake in Windsurf. I guess the people are worth more than the company.
Google made this announcement after talks between Windsurf and OpenAI ended. But this hasn’t slowed down OpenAI. They bought Jony Ive’s device startup for $6.4b. And now, after those Windsurf talks, it seems they’ve concluded that LLMs are rapidly becoming commodities and dead ends for making money. As such, they’re going to move into the consulting business and follow Palantir’s engagement model. I think this is the right move for them, but it will have repercussions on the big consulting houses, like Accenture, Deloitte, IBM, and perhaps McKinsey (which in turn will make these consulting players think about acquisitions). Sounds like Dune!
That being said, in my way of thinking, this move suggests a need for new types of expertise for OpenAI: line managers to scope out the opportunities and projects, vertical expertise to understand the customer, programmers that adhere to SOWs and timelines, just to start. It’s hard to do this well. And it’s expensive!
To add significantly to OpenAI’s bottom line, I bet they’ll acquire more companies to offset their net burn. I’ve read articles suggesting they lost $5b after revenue, which some say was really about $4b. That means they’re spending $9b on year, and it might very well get worse to keep up with their competitors. Maybe that’s why OpenAI opted out of acquiring Windsurf, and perhaps instead, they will focus on Palantir. Why not?
With hopes of rocketing past competitor OpenAI, Elon Musk’s SpaceX announced it will invest $2b in his startup company xAI, which developed their chatbot Gronk. Gronk received high marks from the benchmarking service Artificial Analysis for its performance, but it seems to have earned some low marks after posting racist comments on X last week.
Hoping to strengthen their positioning in the ever-growing and competitive AI real estate, it was reported that Apple is in talks with Anthropic about replacing Siri with a version of Claude. But this could be in the name of negotiating leverage because, given Apple’s secretive nature, why would the press know about these conversations in the first place? Or maybe Apple wants both companies to make up for lost time. Regardless, all moves will amount to billions.
Other notable mentions are Databricks buying Neon for a mere billion, and Snowflake condescending to get Crunchy for a paltry $250m, in their aspiration to offer an alternative to AWS’s RDS and Aurora database platforms.
Going back to my MariaDB days, I’m not sure which is harder: To compete with AWS on databases or make money on ads using WhatsApp. Maybe the database area is harder, but perhaps that’s not Databricks’ intention at all. Perhaps they’re wrapping themselves up to be acquired by OpenAI or Nvidia? While they’re not consulting companies, they do work directly with customers and make money. Who knows. A game is certainly being played.
Then there’s Oracle, which stayed relatively quiet during this yard sale of the rich and mighty, only to watch their stock hit a record closing high last week after disclosing a multibillion-dollar deal with an unnamed customer. Turns out that the customer is OpenAI, which plans to rent 4.5 gigawatts of data center capacity from Oracle. This deal will apparently lead to $30 billion in annual revenue beginning in fiscal 2028.
So, how much does it cost to make money? For them, billions. For me, ingenuity.
-mh