4 on a Monday: ESI to become part of Keysight Technologies; Hexagon and Cadence acquire; and the cost of AI to you and the developer

ESI announced that Keysight Technologies would acquire it in two stages for about EU 913 million. For those keeping track, that’s roughly a 6.5x revenue multiple.

Who? Keysight had revenue of $5.4 billion in 2022, roughly half from commercial communications companies, and the rest split between aero/defense/government and electronics and industrial buyers. It serves 14 of the top 15 top mobile operators, 5/5 of the leading cloud providers, 22/25 of the top aero and defense contractors … you get the idea. Big, Global. Many of the same customers as ESI, or those ESI would love to support. 

Why? In prepared remarks, Keysight CEO Satish Dhanasekaran said that the “acquisition of ESI Group accelerates our strategy of providing software-centric solutions with virtual prototyping and advanced simulation capabilities.”

Why, from ESI’s perspective? ESI CEO Cristel de Rouvray said she is “delighted for ESI to announce this significant milestone with Keysight. Over the past four years, our entire team has been dedicated to transforming ESI Group into a leading actor in the industrial software and simulation sectors. This combination with Keysight … will enable ESI to enter a new phase of its development … an exciting and sustainable future.”

MY opinion? It’s hard going as a small player in the consolidating world of CAE. ESI’s resources for R&D, its ability to market and sell globally, and access to talent were limited by its size. They were doing all the right things to correct structural problems within the company and could eventually have turned to more robust growth. But that will happen sooner with Keysight’s backing, capital, and market access. Keysight gets a valuable software asset that it says will bring its annual software revenue to ~$1.3B or ~23% of total revenue. 

I am quite sad about this —though excited for the ESI team on this next chapter— because it’s another smaller standalone fading from the stage.

You can see Keysight’s investor presentation here: https://s22.q4cdn.com/444849635/files/doc_presentations/2023/keys_esi_20230628.pdf

Hexagon said last week that it had acquired HARD-LINE, makers of mine automation, remote control and mine production optimization technology. The company says HARD-LINE will “greatly enhance Hexagon’s life-of-mine technology and safety solutions, and will particularly complement [its] newest portfolios, HxGN Underground Mining and HxGN Autonomous Mining.” Financial details were not provided, but the acquisition is not expected to impact Hexagon’s earnings.

Cadence Design Systems announced it will acquire Rambus Inc.’s SerDes and memory interface PHY IP business. Rambus’ engineering teams in the United States, India, and Canada will join Cadence to support these products.

Why? Cadence says, “The expected technology asset purchase also brings Cadence proven and experienced PHY engineering teams … further expanding Cadence’s domain-rich talent base.” (Rambus will retain its digital IP business, including memory and interface controllers and security IP.)

Finally … We are starting to see how much AI-enabled applications will cost us. This week, Microsoft said it would add AI to your Office subscription for $30 per month (in addition to whatever you’re currently paying). It calls this assistant (remember Clippy?) Microsoft 365 Copilot. Copilot uses technology from OpenAI, makers of ChatGPT, to comb through public info from the Internet and a company’s internal information, to compile the answers to questions, help build and analyze spreadsheets, create PowerPoint presentations, and more. Microsoft didn’t say when Copilot will be broadly available. More here https://www.engadget.com/microsoft-will-charge-businesses-30-per-user-for-its-365-ai-copilot-153042654.html

For more on the costs of embedding AI in commercial software solutions, see here: https://www.runtime.news/the-price-of-ai/ . Custom AI-specific chips, training algorithms, and ensuring security are all expensive. The linked article says that these costs are slowing down the adoption of AI, as is the need for more computing capacity to train models. Hardware and software vendors, as well as data scientists, are all searching for commercial models that are both profitable and priced to grow adoption.

We’re still feeling our way in this AI-enabled-everything world. How to create value, what to charge for it, how to protect the IP of the creators whose work enables the learning mechanisms … all very unclear. What is more clear is that specific applications, like using AI to supplement CAE runs or generate drawings from models, do offer quantifiable benefits that users will be willing to pay for.

What about you? Are you interested enough in AI to pay for something that has it embedded? How much would you pay for that? To me, $30/month for help in creating PowerPoints isn’t worth it.