AI in charge: how companies are rewriting the rules of business.
OpenAI, Google, Musk, DeepSeek (and more): whoever governs technology can also control business. And no one wants to be left behind.Published in issue 3, May/June/July 2025 of MIT Sloan Management Review Italy
In the last two years, theArtificial Intelligence (AI)has stopped being just a technological frontier and has become the real thingstrategic battlefield of the global economy.The race for innovation is no longer played only on the terrain of new models or platforms, but on that ofcapital integration,of know-how andintellectual property.
In this context, Mergers and acquisitions (M&A) operations are today the privileged lever foraccelerate growth, acquire distinctive technologies and strengthen digital resilience.The quality and scope of acquisitions matter more than quantity: large groups and more dynamic scaleups (defined as companies that have grown for three consecutive years at an annual rate of over 20%, in turnover or number of employees) focus on targets with unique assets - data, patents, talent, infrastructure - to control strategic segments.
At the same time, thecompetitive pressureand the availability of capital fuel a season ofdeal makingwhich involves not only pure tech, but all the industrial chains touched by AI. However,macroeconomic and geopolitical uncertaintyimposes increasingly selective acquisition strategies and growing attention to the creation of concrete synergies and speed of integration.
The two-year period2024-2025thus marks the transition from the era of widespread experimentation to that oflarge strategic maneuvers,where M&A is one of the keys tobuild value and defend leadership in the ecosystemglobal AI, without losing sight of the central role of technology and the ability to innovate.
A new form of acquisition, known asreverse acquihire,in which a large company takes on a significant part of the management team of a smaller company and acquires its technology throughlicensing, avoiding the complications of a traditional acquisition. An example: in August 2024,Googleorchestrated such a $2.7 billion operation withCharacter.AI.
Market growth and investments
In 2024-2025, AI recordedrecord investmentsand an acceleration in its adoption, transforming businesses and sectors. In Italy, the market grew by 58% in 2024, reaching i1.2 billion euros, led byGenerative Artificial Intelligence (Gen AI) and large companies.
Globally, AI tech revenues are growing rapidly:Microsoft,for example, it plans to generate 10 billion dollars annually from AI alone within two years. In the United States, in 2024, AI has absorbed the46.4% of investmentsof Venture capital funds, or 96.9 billion dollars. In Europe, the share was 20% (around eight billion). Thehyperscaler(Amazon, Microsoft, Alphabet and Meta) plan to invest collectively320 billion dollars in 2025in AI and data centers, against 230 billion in 2024.
The major players on the market
OpenAI.Founded in 2015 in San Francisco by a group of leading entrepreneurs and researchers, OpenAI is aresearch laboratory on Artificial Intelligence. Born as a non-profit organization with the mission of promoting safe and beneficial AI, it then adopted a 'capped profit' model to attract investments and accelerate research. Microsoft became its main strategic partner, investing 13 billion dollars and providing advanced cloud infrastructure. In 2025, OpenAI picked up40 billion from Softbank, reaching a valuation of 300 billion.
It has redefined collaboration between research and industry, accelerating the concentration of technological and financial resources and creating new barriers to entry for competitors.
Anthropic.Founded in 2021 in San Francisco by Dario and Daniela Amodei and other former OpenAI researchers, Anthropic was born from strategic divergences with the aim ofsafety, ethics and alignment with human values at the center of AI development.Established as a Public Benefit Corporation, it stands out for its 'AI safety first' approach and for its reputation for prudence in the diffusion of its models, such as those of the Claude family. The company raised more than that18 billion dollarsfrom partners such as Amazon (8 billion), Google, Salesforce and Lightspeed venture partners, which led a 3.5 billion round in 2025, bringing the valuation to61.5 billion.Amazon web services (AWS) is the primary cloud provider. In addition to the development of advanced models, Anthropic invests in research on interpretability, scalable supervision and social impacts, positioning itself among the global leaders in responsible and trustworthy AI.
Perplexity AI.It was born in 2022 in San Francisco by Aravind Srinivas (former OpenAI employee), Denis Yarats (formerly Meta), Johnny Ho (formerly Quora) and Andy Konwinski (formerly Databricks), with the aim ofrevolutionize online researchthrough a conversational engine based on Gen AI.
Unlike traditional engines, Perplexity interprets questions in natural language, consults the web in real time and generates synthetic answers accompanied by verified sources. It integrates advanced models such as Gpt-4, Claude 3, Mistral Large, Llama 3 and a proprietary model, taking advantage of the Retrieval augmented generation. The platform is 'freemium': the free version uses proprietary Claude 3 Haiku and Large language models (Llm), the Pro offers access to more advanced models.
In 2024 it surpassed the10 million monthly users.In May 2025 it is in negotiations for a round of500 million dollars(valued at 14 billion), led by Accel and with investors such as Nvidia, Jeff Bezos, Databricks and Yann LeCun.
xAI. Elon Musk's startup, founded in 2023, was created to develop advanced AI systems and to 'understand the true nature of the universe'. In just a few years it has positioned itself among the main world laboratories, in direct competition with OpenAI and Anthropic.
The flagship product is Grok, a generative chatbot initially integrated on X (formerly Twitter) and, from 2025, also on Telegram thanks to a 300 million dollar partnership. In March 2025 xAI acquired This merger allows xAI to accelerate the development of its models by leveraging real-time social data, global distribution and synergies between technical teams.
After the acquisition of X, xAI wasvalued at around 80 billion dollars. Musk remains the largest shareholder; investors include funds such as Sequoia, Fidelity and Vy Capital. Negotiations for new rounds are underway in 2025 which could bring the valuation up to $120 billion.
Gemini [Google]. It is the family of next-generation Gen AI models developed by Google, thanks to the collaboration between DeepMind and Google Research. Presented at the end of 2023, it represents the most advanced and flexible AI system ever created by Google.
Designed to understand, generate and combine text, images, audio, video and code, Gemini uses deep neural networks,architecturestransformersereinforcement learningto offer high performance in language understanding, code generation and data analysis. It is integrated into Search, Workspace, Android, Vertex AI and accessible via apps and APIs.
His strength is advanced reasoning and the ability to solve complex problems with complex chains of thought. The model is optimized for efficiency and scalability, adapting to both data centers and mobile devices, and represents the centerpiece of Google's AI strategies for the next decade.
Mistral AI. It is a French startup created in 2023 in Paris, by Arthur Mensch, Guillaume Lample and Timothée Lacroix, former DeepMind and Meta researchers. Its goal: to democratize AI through efficient, transparent, and often open source models (i.e., the source code is made freely available for modification and redistribution).
The company develops both open source and commercial LLMs, characterized by architecturestransformersemixture of experts, which guarantee high performance with fewer computational resources. The models are also suitable for regulated sectors such as banking and healthcare.
In June 2024, Mistral closed a 600 million euro round (valued at 6 billion euros) with investors such as Microsoft, Nvidia, Salesforce, General Catalyst and others, positioning itself among the main European AI players.
DeepSeek. It is a Chinese company, founded in Hangzhou in 2023 by Liang Wenfeng (co-founder of High-Flyer, main financier). The company has quickly positioned itself among the world's leading AI laboratories.
DeepSeek has developed a range of extremely efficient and low-cost Llms (DeepSeek-Llm, V3, R1), with open weight models (in these cases, the model provides access to the parameters of thetraining, but often excludes data fromtraining, the source code and the methodology) and partly open source, which allow transparency and free, or very cheap, use compared to Western competitors. The company has optimized training costs thanks to techniques such asmixture of expertsand to the use of less powerful hardware, bypassing restrictions on exports of AI chips.
In 2025, the DeepSeek app reached first place on the US App Store, surpassing ChatGpt in downloads. The success of the models had a significant impact on the global market, contributing to the collapse of Nvidia shares (which lost 17% of their value on Monday 27 January 2025, but subsequently saw a strong recovery: on 29 May they were only 2% below the price on Friday 24 January). DeepSeek also stands out for its meritocratic culture and openness to talents from different backgrounds. DeepSeek is self-financed through High-Flyer and has not raised capital from external venture capital. SecondForbes, theestimated valuation ranges from 1 to 10 billion dollars, confirming it among the top five AI laboratories in the world.
Meta (Llama). Meta has undertaken a massive and targeted investment strategy in the development of the Llama family of AI models, with the aim of consolidating its position in the open source and Generative AI sector.
Investments between 60 and 80 billion dollars are expected for 2025 in infrastructure, data centers and over 1.3 million GPUs, fundamental for the training and scalability of Llama 4.
The 'Llama for startups' program encourages adoption among startups by offering cloud support and credits. Meta is aiming for a hybrid model: free access for non-commercial users and SMBs, paid plans and licenses for large enterprises and hyperscalers, as well as partnerships with cloud platforms such as Aws. Llama 4 represents a technological leap compared to Llama 3, with 10 times the computational power and advancedself-supervised learningereinforcement learning with human feedback.Meta expects revenues from Llama of between £2 billion and £3 billion as early as 2025 and between $460 billion and $1.4 trillion by 2035. Despite success in terms of adoption (over one billion downloads), Llama has faced delays and criticism over the transparency of benchmarks. Meta is aiming for long-term returns, both in terms of innovation and new revenue streams.
Alibaba. In 2025, Alibaba announced a record investment of 380 billion yuan (about $53 billion) in AI and cloud for the next three years, the largest in its history. The investment is mainly intended to strengthen infrastructure, data centers and computational capacity for training advanced AI models.
Alibaba aims to consolidate global leadership in cloud services and AI, also focusing on Artificial General Intelligence (AGI) as a long-term strategic objective. The company presents itself as a point of reference for companies that develop and implement AI solutions.
The commitment to AI has already contributed to significant revenue growth, particularly in the Cloud Intelligence division (up 11% year over year) and in AI sales, which have recorded triple-digit growth for six consecutive quarters. Table 1 summarizes the key data for each of the nine AI companies analyzed.
Recent developments
In the month of May, several interesting news related to the AI world were published in the international press. A group of startups (Pip Labs, Vermillio, Created by Humans, ProRata, Narrativ and Human Native) are buildingtooland platforms where writers, publishers, music studios and film producers can receive compensation if they allow the use of their content for training AI models. These platforms have received $215 million in funding since 2012. It is expected that thethe AI licensing market will grow from 10 billion in 2025 to 67.5 billion in 2030.
Microsoft, however, has made the AI models of Elon Musk's startup xAI available to its cloud customers, in the latest sign of a cooling in Microsoft's relationship with OpenAI, inventor of ChatGpt.
Google, on the other hand, plans to add features to its search engine that make it similar to a chatbot AI, at a time when it is busy competing with competitors like OpenAI. US-based Google Chrome and Google Search users as of last month can enable an AI mode that provides a ChatGpt-like 'conversational' interaction, rather than the traditional list of links. With this move, Google is moving further away from its free, ad-supported model. The standard “AI pro” subscription costs $25 per month.
On the opposite front, OpenAI acquired, for 6.4 billion dollars, the startup founded by Sir Jonathan Ive, who had been Apple's Chief Designer and who had played a fundamental role in the creation of the first iPhone in 2007, working closely with Steve Jobs. Sir Ive left Apple in 2019 after working for almost 30 years at the company. OpenAI's goal is to create new products aligned with the AI era.
Four of the oldest industrial groups in Europe (Schneider Electric, Siemens AG, ABB and Legrand) have seen their stock market valuations increase by more than 150 billion dollars overall, thanks to the growing demand fordata centerslinked to the explosion of AI. These groups continue to supply electrical equipment for residential and industrial customers, but data centers represent the fastest growing portion of their revenues.
Future scenarios
AI enters a phase of maturity where growth will no longer be driven only by hype, but byability to integrate into processescore,inscalability of vertical solutionsand molesmeasurable economic returns.In the coming years, we will see a progressive convergence between large global platforms, innovative startups and industrial operators, with an increase in strategic partnerships and M&A operations aimed at consolidating critical assets – data, chips, proprietary models and talent.The adoption of AI agents and solutionsedgewill accelerate the spreadin sectors such as Manufacturing, Healthcare, Energy and Defense, while demand for hardware and cloud infrastructure will continue to grow.
The main uncertainties. The competitive and regulatory framework remains highly fluid. The fragmentation of regulations between the United States, Europe and China requires companies to adapt products and strategies to different markets, increasing compliance costs and risks. Competitive pressure leads to ainvestment rushunprecedented, but also exposes it to the risk of overvaluation of assets and possible sectoral 'bubbles'.
On the technological front, the speed of innovation makes it difficult to predict which models, chips or architectures will truly dominate the market in the coming years, while thetalent shortages and data security.
What to monitor?Management and investors will have to pay attention to several key factors: the ability of companies to quickly integrate acquisitions and generate concrete synergies; the evolution of international regulations and the risk of restrictions on technological exports, privacy and the use of data; the emergence of new 'AI first' business models and the scalability of vertical solutions, which will be able to make the difference between those who create real value and those who remain trapped in experimentation; the sustainability of investments in infrastructure, with particular attention to energy, supply chain and cybersecurity.
New organizational and leadership models.Onenew generation of leadershipis taking shape, the‘AI first’.The ability to orchestrate digital ecosystems, enhance human-machine collaboration and integrate AI into decision-making processes becomes central.Successful organizations invest in widespread training,adopt agile and cross-functional models, and create roles dedicated to AI governance and risk management. The focus shifts from technical implementation alone to the creation of a corporate culture that knows how to exploit AI to innovate, redesign processes and anticipate market developments.
In summary,the future of AI will be guided by those who know how to balance speed of execution, integration capacity, attention to compliance and strategic vision.Companies that invest in leadership, governance and partnerships will be able to transform innovation into sustainable competitive advantage.