The Economist: Joint business ventures in AI and carmaking on the rise as takeovers on the decline

No firm is an island. All strike contracts and compete with others.

Conversely, when bosses decide a particular relationship would be better governed by fiat, one firm may acquire another. Between these poles are plenty of ways for firms to combine capital, knowledge or other resources, without fully tying the knot.

Such in-between arrangements are winning favour across the economy, from tech and artificial intelligence (AI) to carmaking and energy. While corporate takeovers stalled in 2023, the number of joint ventures (JVs) and partnerships jumped by 40 per cent, according to Ankura, a consultancy.

They are especially popular in areas of rapid technological change and in places given to protectionism, which these days afflicts rich and poor countries alike.

As barriers to commerce rise, high interest rates continue to bite and regulators bridle at takeovers, such liaisons are becoming the go-to ways to enlarge a business empire, as the recent actions of firms including Disney, Ford and Microsoft illustrate. Call it the age of the quasi-merger.

When the scope of co-operation is clear, firms often choose to share ownership of a separate entity through a JV. In February Disney announced a new sports-streaming service bringing together its own ESPN network with the content of two entertainment rivals, Fox and Warner Bros Discovery. Weeks later it announced a similar move in India, joining forces with Reliance, one of India’s biggest conglomerates, in an $8.5bn deal.

Many recent constructs are fuzzier. Microsoft has forged partnerships with the world’s hottest makers of AI models: OpenAI of San Francisco, Mistral of Paris and, this month, G42 of Abu Dhabi. Investments by the world’s most valuable company will give it minority stakes in Mistral and G42. After backing OpenAI to the tune of $13 billion, it holds a non-controlling interest in the ChatGPT-maker’s for-profit subsidiary.

Last February Ford, an American carmaker, announced a partnership with CATL, a Chinese battery giant, to build a $3.5b battery factory in Michigan. The Chinese firm would provide the know-how via a licensing deal, but not own a stake in the project.

If America manages to force TikTok’s Chinese owner, ByteDance, to sell the social media app under pain of a ban in America, it would probably end up in the hands of a consortium of American firms.

Quasi-mergers are not new. Firms have long teamed up to manage project costs, new technologies and manufacturing-obsessed governments.

This year Renault of France and Nissan of Japan are celebrating a quarter-century of the sector’s biggest alliance, which Mitsubishi, another Japanese firm, joined in 2016. CFM International, which is jointly owned by GE Aerospace, an American industrial firm, and Safran, a French one, has been making aircraft engines since the 1970s.

In the 1990s, notes Melissa Schilling of New York University, companies scrambled to form new partnerships in order to capitalise on the new technology of the day — the internet. Developing countries, including China, have long obliged foreign firms looking to take advantage of cheap labour and vast markets to hand over technology through JVs with local partners.

Bye bye, tie-ups. Hello, team-ups

Today’s more complicated world is leading to more complicated arrangements.

One hotspot is, once again, carmaking. The industry is being remade by the shift from combustion engines to electric vehicles (EVs)—and by fears of deindustrialisation as Chinese carmakers dominate the market.

Last October Stellantis, formed by the merger in 2021 of Fiat Chrysler and PSA Group, owner of Peugeot and Citroën, announced the purchase of 20 per cent of Leapmotor and the establishment of a JV to build and sell the Chinese EV-maker’s cars abroad. (Stellantis’s biggest shareholder also part-owns The Economist’s parent company.)

The next month Renault and Nissan ratified a new, looser pact with more equal cross-shareholdings. In March Nissan and Honda, another Japanese rival, said they were exploring a strategic partnership to develop EVs.

Many of the new carmaking ventures are not about making cars — at least not directly.

Last year Stellantis bought nearly 20 per cent of McEwen Copper, a small miner, as part of a deal (which also involves Rio Tinto, a giant one) to extract the red metal in Argentina. That copper may eventually make its way to Kokomo, Indiana, where Stellantis owns 49 per cent of two battery factories being built with Samsung SDI, a South Korean battery firm part-owned by the electronics giant of the same name. Stellantis is also part of IONNA, a JV among seven carmakers that plans to build 30,000 charging stations in America.

Titans of digital technology are building similarly intricate webs of co-operation.

In contrast to arrangements between lumbering carmakers, where the main rationale of limited co-operation is to spread costs, the AI deals have more to do with antitrust authorities’ conviction that big tech is already too big.

In March Amazon said it had invested $4b in Anthropic, securing access to its Claude 3 model for its customers and crowning itself the model-maker’s “primary cloud provider for mission-critical workloads”. Alphabet has promised Anthropic up to $2b and also crowed that the startup uses its cloud infrastructure.

Microsoft, AI’s most ambitious dealmaker, is well aware of the dangers posed by stickier red tape. Its full-blown acquisition of Activision Blizzard, a video-game developer, took nearly two years to complete and was nearly derailed by trustbusters.

The software behemoth first began working with OpenAI in 2016; $13b later and it is integrating OpenAI’s models into its consumer and enterprise products. In the shadow of turbulence at the startup, which in November led to the swift firing and re-hiring of its boss, Sam Altman, Microsoft has begun to spread its bets.

The $16 million investment in Mistral, announced in February, may be tiny but helps bring France’s best shot at becoming an AI superpower into Microsoft’s orbit.

Even more creatively, in March Microsoft stunned the tech world with a “no-deal” deal whereby senior employees at Inflection AI, another model-maker, decamped to Microsoft. The startup’s other investors are reportedly being compensated by an unusual licensing agreement. (One of Inflection’s founders sits on the board of The Economist’s corporate parent.)

Microsoft’s $1.5b G42 transaction is half partnership and half high-stakes diplomacy — the deal came with hope of closer co-operation on AI between the American and Emirati governments.

The success of the quasi-merger wave is hard to predict. Although alliances pass the desks of regulators and politicians more easily than takeovers, they can still come unstuck.

Last year American Airlines and JetBlue, another carrier, ended their alliance on America’s east coast after being sued by the Department of Justice.

Disney’s new sports venture is getting a close look from antitrust referees. Earlier this year the Federal Trade Commission, another American antitrust cop, opened an inquiry into the AI deals. European and British regulators are making similar noises.

Cross-border deals, in particular, tread a narrow path. Teaming up in emerging economies has always needed careful management, lest politically connected local firms turn on their foreign partners or an entire jurisdiction becomes uninvestable.

The turn away from free markets in the West has, to some extent, globalised this political uncertainty. The flexibility inherent in a partnership or JV but absent from a full-on takeover could make such structures more politically acceptable.

But even transactions crafted to avoid tripping wires can come under scrutiny. Although CATL’s partnership with Ford does not involve an equity investment by the Chinese firm, that hasn’t stopped American lawmakers from calling for close scrutiny of the deal.

Perhaps the biggest threat to the new-fangled partnerships is the partners. Aligning firms’ incentives is notoriously tricky. The crucial fine print of quasi-mergers is kept secret, giving shareholders little insight into what bosses have actually agreed to.

No wonder disputes are ten a penny. In March an American judge ruled that Walmart could end its partnership with Capital One after the supermarket and credit-card provider fell out over the terms of their agreement.

A fraught game of linguistic chess has the oil industry on the edge of its seat. ExxonMobil is warring with Hess, a smaller rival, over what should happen to their JV in Guyana should Hess sell itself to Chevron, Exxon’s rival American supermajor.

A global slowdown in EV sales will put stress on carmakers’ new corporate constructs (Ford has already scaled back plans for its battery plant in Michigan).

Being a novel technology, AI raises novel questions about matters such as safety or copyright that do not necessarily lend themselves to shared decision-making. Quasi-mergers are here to stay. Many may prove only quasi-successful.