Positioning and politics in the world of Artificial Intelligence

Or: what can we learn from the story of Mercedes and Bosch?

figure: electric SMART 451 ED, an early attempt to build (great!) electric cars and learn battery manufacturing by Mercedes in the 2010´s. Current European battery production could be overly reliable on Asia for raw materials. Finding alternatives for raw materials needed in battery cell production that are more commonly available is of the essence.

Mercedes has been a premium car builder for years, with as main competitive advantage very high quality of drivelines and high protection levels of vehicles for the people inside. The long livedness of the cars is legendary, and it is not seldom to find Mercedes cars with more than a million miles driven.

For years, Mercedes had its competitive advantage in their robust drivelines, the engines, gear boxes and differentials, which were of utmost quality and required huge investments to design, test and build. Mercedes reportedly invest 8.5 Billion Euro in R&D every year for its car and truck divisions (Statistica, 2023).

For electronic parts, the designated partner of Mercedes is Robert Bosch AG, a global player in car electronics. Bosch has delivered most critical parts for all operational and security functions for mercedes for decades.

But the last ten years, this cooperation and symbiosis started to show cracks. Electric cars entered the stage, and completely turned the business competitive advantage for Mercedes upside down. For developing electric cars, Mercedes became fully relying on external knowledge for their most important and distinguishing feature: the drive trains. Batteries and electro-engines are not new inventions and have been around for centuries (!). In automotive, the typically became available from organisations as Bosch AG, Panasonic and others. All of a sudden the cars´ drivetrain was not the part that required billions of Euro investment anymore, but became something that could be ordered with Bosch & Co. Tesla smoothly hopped into the space with the help of Asiatic battery producer Panasonic. At that time, high-tech battery development and production in Asia had already gained a head start, largely unanswered by the west, and this lead might be difficult to regain.

Last month the stock price of Mercedes dropped with a few percent. Reason: external supplier Bosch had delivery issues with its 48V on-board batteries and thousands of Mercedes S-classes could not been build and delivered to the market. An external, single point of failure. Even though Bosch was a victim of issues in their own supply chain, the company lives well as they are not depending on one client in the receiving end. Such asymmetric dependencies are difficult to manage over time. Mercedes announced it will build their own batteries and electrical engines in the future. It aims at regaining control over their competitive advantages and production capabilities. Problematic for that plan is that Europe has some battery factories for assembling battery packs with high complexity and safety, but that the know how of cell design and manufacturing is largely Asian.

Now relate this to the situation in the world of Artificial Intelligence. We have a company, Microsoft, that has a good hold of the cloud market, where many organisations put their data and use cloud-services to create business value. One could say that the competitive advantage of Microsoft is in their «drivetrain»: cloud and cloud services.

Then, they find that one of their competitors, Google, is ahead in the race to build value added services on cloud with AI. Google develops very advanced artificial intelligence, which utilised enormous amounts of data available on the internet to create services that are difficult to duplicate for those without access to such. The new services build upon available external data to deliver services that are complementary to a clients own data, which can augment this «local» data with information about the world and how to communicate about it. Microsoft becomes weary of loosing out and starts investigating their options. They need a Bosch or Panasonic to help them fill the gap.

And this Robert Bosch they found in OpenAI. External, small, cool and with a service that was not easily replicated. Microsoft invested heavily in this small startup and made their product, which competed with Googles BERT and other models, into their main cloud services. The small partner launches a demo of their product online and nearly immediately gets world famous. Microsofts stock soured and their strategy seemed to work, none of the competition was prepared and many red-flags were raised all over the competitive landscape.

All looked good. Until last week. Microsoft experienced its «Mercedes» moment, when Sam Altman addressed several of the issues with Generative AI, while simultaneously launching new functionality addressing those issues. He boldly stated that these features would only be available through OpenAI directly. This meant in practice that Microsoft was stuck with an older version of their product, not addressing many of the criticisms of the Microsoft OpenAI functionality. And on top of that they do not have control over this smaller partner while owning only 49% in OpenAI. So the same situation appears as in the automative market, where established AI Big Tech is depending on partners that are small and previously regarded «innocent» or at least «controllable». All of a sudden these become the biggest threat to operations as a single point of failure.

At the developer Days by OpenAI (November 6th, 2023), Sam Altman stiffly and clearly memorised the whole presentation of all new, cool features in best Silicon Valley style. But then something happened in the middle of the seance: Microsoft boss Satya Nadella

makes an appearance. Where we just learned that OpenAI seemingly went alone on building next-gen features of their product. Something felt sincerely wrong that day. Why was Satya on stage telling us all was good and Microsoft and OpenAI loved each other? No smoke without fire, was what we discussed that week. But what was going to happen? On Friday November, 17th 2023, the answer to that question came, Sam Altman was fired. In an utterly direct blog on OpenAIs web page it was communicated that «Mr. Altman’s departure follows a deliberative review process by the board, which concluded that he was not consistently candid in his communications with the board, hindering its ability to exercise its responsibilities. The board no longer has confidence in his ability to continue leading OpenAI.».

With other words, Microsoft is experiencing its Bosch moment and aims at regaining control of the future of their business. The leadership change made Microsoft stock drop (a few percentage points), but compared to the enormous gains earlier this year based on this cooperation, this might not be something their leadership will be worried too much about for now. The most important will be to secure control over their main competitive edge, the one that brought them back at the table again. And having control over their own capability (the model) and the supply train (training such models) is key. So the announcement that Microsoft builds their own chips, together with Sam Altman leaving their achilles heel, should not come as a surprise, but should be seen as logical conclusion on processes we have followed for a few months now.

Robert Engels (CTO AI, Capgemini) with inputs and feedback from Wout van der Kooij (retired member of the Board of Directors of Varta AG and Linde AG´s Gas Division Board)

Handelsblatt (2010): https://www.handelsblatt.com/unternehmen/industrie/daimler-und-bosch-bei-den-traumpartnern-kriselt-es-heftig/3530336.html

Mercedes/Bosch (in German) — Stuttgarter Nachrichten (2023): https://www.stuttgarter-nachrichten.de/inhalt.lieferausfaelle-dicke-luft-zwischen-mercedes-und-bosch.34df75bb-5bac-49a1-9e33-82bea5a9aeb0.html

Microsoft introduces its own chips — Reuters (Nov 15, 2023): https://www.reuters.com/technology/microsoft-introduces-its-own-chips-ai-with-eye-cost-2023-11-15/

R&D expenses Mercedes Cars — Statistica (2023): https://www.statista.com/statistics/346985/research-and-development-expenses-of-mercedes-benz-cars/