Nvidia’s market value has risen past $3 trillion, overtaking Apple as the world’s second-most valuable company by market capitalisation. Microsoft remains top dog, but the Nvidia-frenzy appears to have built inexorable momentum, perhaps making its ascension inevitable.
Tech giants like Google, Microsoft and Meta are heavily investing in Nvidia’s AI chips, which comprise 70%-95% of the market for training and deploying models, like OpenAI’s GPT. And yet, seemingly at every conference and in every quarterly report, Nvidia continues to exceed these heady expectations.
CEO Jensen Huang announced the next generation of AI chips, “Blackwell,” and tested future “Rubin” processors at the Computex conference earlier this June. The announcement came only three months after their last launch, evidencing Nvidia’s aggressive approach to consolidating its grip on the AI chip market.
Nvidia’s push for market domination has been met with increasing resistance. AMD’s Instinct MI300X AI chip has been adopted by Microsoft for its Azure cloud services, whilst Intel’s Gaudi 3 AI accelerator is being marketed as a cost-effective alternative to Nvidia’s H100. Major cloud providers like Amazon, Google, and Microsoft are also developing their own AI chips - Nvidia’s monopoly may not be long for this world.
But the question practically raises itself: Is this truly competition?
Most AI start-ups and research labs depend on Big Tech for computing infrastructure and market access, licensing and rebranding AI models created by these tech giants. Big Tech has accrued significant advantages through platform monopolisation and their data-harvesting business model, defining the incentive structures for AI research and development. Meanwhile the market is increasingly consolidated, both vertically and horizontally.
AI start-ups are braced for a wave of mergers and acquisitions as many are facing financial difficulties. Talent and AI-standard computing costs force smaller firms to burn through funds and Big Tech are swallowing struggling start-ups whole. 2024 has seen 105 AI and machine learning deals in the US worth $35 billion, already outstripping 2023’s total.
This figure will balloon once Apple and OpenAI announce an expected partnership which would inter alia integrate ChatGPT into the iPhone’s operating system. The deal could make ChatGPT a core feature on every iPhone by improving Siri or even becoming its defining feature.
The Big Tech firms have leveraged their competitive advantage, strengthened it in the process, and entrenched their economic and political standing. Competing in AI is the preserve of but a few monopolists.
Democracy, culture and individual agency are threatened by this concentration of power. Centralised control over critical technology is no good for politics. Take Cambridge Analytica. Personal data, harvested without from millions of Facebook users, was used to influence voter behaviour in the 2016 US presidential election and the Brexit referendum.
Grave concerns extend as far afield as adult entertainment. An AI brothel is set to open in Berlin. Patrons will be able to talk to and physically interact with AI ‘dolls’. Dated stereotypes about pleasure may well define these dolls as well as broader AI chatbots. An undeniable risk of addiction, with all its usual accompanying mental harms, emerges here - particularly for lonely people, notably men. AI companion apps have reached 225 million downloads on the Google Store. The industry is enjoying an unrestrained, socially deleterious boom.
The monopolistic development and deployment of AI creates overreliance on a handful of computer systems. Malfunctions or security breaches would have far-reaching consequences difficult to forecast.
The concentration of power stoppers competition and stifles innovation. Public interests are above the promises and demands of Big Tech, who must be made to reckon with this. Countries, notably the US, must regulate to promote accountability and competition in a manner accommodating the vicious lobbying this will precipitate.
Meta already broke its record for lobbying in the first quarter of 2024. In 2023 Microsoft invested £2.5 billion in cloud infrastructure, blunting a Competition and Markets Authority investigation into the cloud market. Prime Minister Rishi Sunak, in attempts to cultivate a British AI sector, praised the move.
So what can be done?
Improved regulation, given its presently feeble state, is despite concerted lobbying eminently possible. The way (read policy) is clear, but the will (read politics) is lacking. Leaders eager to share in the boom are fearful of offending its key players — AI is having its moment and no state is, as yet, willing to saddle it.
Image: Jernej Furman
Comments