About Mistral AI

• The company was co-founded by former employees of Meta and Google’s DeepMind, Arthur Mensch, Charles Gorintin, Guillaume Lample. They develop AI models intended to compete with the latest innovations such as OpenAI’s GPT-4, Anthropic’s Claude 3, and Meta’s Llama 3.
• Mistral AI has released pre-trained and finely-tuned open-source models, such as Mistral 7B, Mistral 8x7B, and Mistral 8x22B, under the Apache 2.0 license, which allows unrestricted use and reproduction, except for attribution. Its most advanced models, like Mistral Large, are proprietary developments designed for transformation into API-accessible products.
• Companies can access Mistral Large via an API, with fees based on usage volume. Mistral AI also offers a free chatbot named Le Chat. Additionally, Mistral AI has partnership agreements with Microsoft Azure, a minority shareholder in the company.

Investment details

American VC fund General Catalyst led the fresh Series B round. Previously, it has already invested in the startup. According to the Financial Times, Mistral AI has raised €468 million in equity capital and €132 million in debt (approximately $500 million and $140 million, respectively). The long list of investors also included Lightspeed Venture Partners, Andreessen Horowitz, Nvidia, Samsung Venture Investment Corporation, and Salesforce Ventures.

Other investors include Belfius, Bertelsmann Investment, BNP Paribas, Bpifrance (through the Digital Venture fund), Cisco, Eurazeo, Headline, venture fund Hanwha Asset Management, IBM, Korelya Capital, Latitude, Millennium New Horizons, Sanabil Investments, ServiceNow, and SV Angel.

After this round of financing, the value of Mistral AI reached $6 billion.

• The capital was transferred to a legal entity called a special purpose vehicle, or SPV, associated with the OpenAI Startup Fund: OpenAI Startup Fund SPV II, L.P. SPVs allow multiple investors to pool their resources and make investments in a single company or fund. In the VC sector, they’re sometimes used to invest in startups that don’t fit a fund’s strategy or that fall outside a fund’s terms. SPVs can also be marketed to a wider range of non-institutional investors.
• In February this year, the OpenAI Startup Fund already raised a $10 million capital through an SPV. This time, the paperwork was submitted on April 25, and mentions Ian Hathaway, the OpenAI Startup Fund’s manager and its sole partner.

About the OpenAI Startup Fund

The OpenAI Startup Fund is a venture fund related to OpenAI, but technically separate from it that came under scrutiny last year, as it turned out that for a long time it was legally controlled by the director of OpenAI, Sam Altman. Although it was positioned as an ordinary corporate venture fund, Altman raised capital for the OpenAI Startup Fund from external partners with limited liability, including Microsoft (a close partner and investor of OpenAI), and had the right to a decisive vote in the fund’s investments.

In April, Altman transferred formal control of the OpenAI Startup Fund to Ian Hathaway, which was previously an investor in venture capital firm Haystack and has played a key role in managing the fund since 2021.

As of last year, the OpenAI Startup Fund had $175 million in commitments and held $325 million in gross net asset value from its incubator program called Converge. It has also supported several startups, including Descript, a collaborative multimedia editing platform valued at $553 million last year, language learning app Speak, an AI-powered note-taking app Mem, and IDE platform Anysphere.

The field of AI is booming and becoming very profitable, but at the same time it raises concerns about biases, privacy, and even the future of humanity. Creators of the bill say that it will make the technology more human-centered.

Advanced position

• According to BBC, the law puts the EU at the forefront of global efforts to tackle the dangers of AI.
• China has already introduced a number of AI laws. In October 2023, US President Joe Biden signed an executive order requiring artificial intelligence (AI) developers to share security findings with the US government. But the EU went even further.
• The EU’s AI law is the world’s first and only set of mandatory requirements to mitigate risks, according to Enza Iannopollo, principal analyst at Forrester. She added that this would make the EU the “de facto” world standard for reliable AI, leaving all other regions, including the UK, to ” play catch-up”.
• The UK hosted an AI safety summit in November 2023, but is not planning legislation along the lines of the AI Act.

How the AI Act will work

• The main idea behind the law is to regulate AI based on its ability to harm society: the higher the risk, the stricter the rules.
• AI applications that pose a “clear risk to fundamental rights” will be banned, for example some of those that involve the processing of biometric data.
• AI systems considered as “high-risk”, such as those used in critical infrastructure, education, healthcare, law enforcement, border management or elections, must meet strict requirements.
• “Low-risk” services such as spam filters will face the lightest regulation – the EU expects most services to fall into this category.
• The law also creates provisions to combat the risks posed by the systems behind generative AI tools and chatbots, such as OpenAI’s ChatGPT.
• This will require manufacturers of some so-called multi-purpose AI systems that can be used for multiple tasks to be open about the materials used to train their models and comply with EU copyright law.

Copyright laws

• The copyright provisions were one of the most lobbied parts of the bill.
• OpenAI, Stability AI, and GPU giant Nvidia are among a handful of AI companies facing lawsuits over their use of data to train generative models.
• Some artists, writers, and musicians argue that the process of “harvesting” vast amounts of data, potentially including their own work, from virtually all corners of the Internet violates copyright laws.
• The bill needs to go through several more stages before it officially becomes law.

Lawyer-linguists, whose job is to check and translate laws, will scour its text and the European Council — composed of representatives of EU member states — will also need to endorse it, though that is expected to be just a formality.

Microsoft, Snap, eBay, and PayPal have cut hundreds or thousands of job positions since the beginning of January, according to Layoffs.fyi, an analytics platform that tracks employee turnover in the industry. Overall, 138 tech companies have laid off staff this year.

These losses are lower than in early 2023. Throughout last year, 263,000 jobs were cut in the tech sector. However, a number of companies, including Amazon, Microsoft, Meta, Google, and Spotify, have announced their plans for optimization this year.

Why does it happen?

Analysts comment that the latest wave of layoffs indicates that companies are reallocating their resources to invest in new areas, such as generative AI, and demonstrate to shareholders that they continue to adhere to cost discipline.

Last year, massive cost-cutting in tech companies occurred in tech companies after they realized that the transition to a digital lifestyle did not last longer than COVID-19. During the pandemic, these companies would actively hire employees, expecting that the change in people’s behavior would be permanent.

However, when life began returning to its previous form, it turned out that there was no longer any need for such large teams. Therefore, they had to cut costs in different ways, in particular by reducing staff numbers.

This year’s layoffs seem to be more strategic than seasonal, said Daniel Kim, associate professor of management at Columbia Business School. “In 2022 and 2023, there was “downsizing” of the staff after the pandemic, i.e., the reduction of jobs that were considered redundant. But this year’s cuts are happening simultaneously with “active hiring” in new areas.

This trend has not bypassed Ukraine either. Recently, Grammarly, a platform for improving written English, announced the layoffs of about 230 team members worldwide, with 37 being from Ukraine.