Qwant, a search engine from France committed to user privacy, and Ecosia, a non-profit search engine from Berlin that channels ad revenue into tree planting and environmental initiatives, are collaborating on a venture to establish a dedicated European search index.
This alliance aims to enhance the innovation capabilities of both search engines—particularly in the realm of generative AI—while decreasing their reliance on the search indexes currently supplied by technology behemoths Microsoft (Bing) and Google. At the moment, both companies utilize Bing’s search APIs, with Ecosia also depending on Google’s search data.
Increasing API expenses serve as a significant incentive for both organizations to reduce their reliance on Big Tech, especially since Microsoft dramatically raised costs for Bing’s APIs last year.
While Ecosia and Qwant will maintain some reliance on Bing and Google, they are determined to diversify the technology underpinning their services by creating their own search index. This strategy is expected to lower operational expenses and enhance their product development efforts as generative AI technologies continue to gain prominence in various digital consumer services.
Both search engines have started incorporating generative AI features. Users can anticipate further developments in this area, although they do not plan to create their own AI models. Instead, they will continue to utilize API access to the large language models (LLMs) of prominent platforms to support these enhancements.
The duo is also receptive to collaborations with other European firms interested in achieving greater technology sovereignty—positioning them as potential clients for the search index, as they aim to license access through an API. They have also expressed openness to other partnership opportunities, as explained to TechCrunch.
“We are ready to engage with anyone who shares our vision,” stated Qwant’s CEO Olivier Abecassis. “However, we also intend to secure investment commitments from our current shareholders.”
“We are confident that we will sustain our operations in the coming years, and our investors are keen to support our rapid progress,” he continued. “We will also engage with investors to facilitate development and encourage partnerships. The goal is to advance as swiftly as possible.”
AI Generating Risks and Opportunities
The rise of AI is instilling a dual sense of urgency in both organizations as it unveils a spectrum of new possibilities and threats.
“The advent of AI tools has shifted the demand for a search index,” noted Ecosia CEO Christian Kroll. “Both Bing and Google have become increasingly hesitant to share access to their indexes. As a search engine, we require an index. This is part of our motivation to ensure our own access.”
“Additionally, we find ourselves in a unique position to utilize such an index to create a transformative experience—leveraging generative AI to develop an innovative approach—and we want to avoid limitations on employing this technology.”
Kroll also highlighted the growing regulatory environment in Europe that is aimed at promoting domestic tech innovation, bolstering the bloc’s strategic autonomy, as a further incentive to invest in a proprietary search index at this time.
“The opportunity landscape has improved significantly,” he stated. “With the [EU’s] Digital Markets Act, for the first time, ‘click and query data’ will be shared among other search engines, so we now have access. The terms of access to platforms have also transformed. While we have contemplated this for a long time, now is the optimal moment to execute the plan.”
“For a significant generative AI user experience, we need both access to LLM models and search technology,” Abecassis added.
The synergy between generative AI models and real-time data sourced through search queries is pivotal in enhancing the utility of search products, he argued.
“We believe that merging both elements will redefine the user experience of searching,” he noted. “Search and generative AI are not interchangeable but can mutually benefit each other, creating a unique blend.”
“Google has opted to keep two distinct products without integration, which is understandable given its longstanding business model. However, we anticipate a convergence between these technologies in the future, which we aim to explore. To achieve this, any market participant will need access to search technology—that’s the service we want to bring forth.”
Towards a European Perspective
The new joint venture, named European Search Perspective, is being established with equal ownership stakes (50:50). Notably, EUP is their selected acronym, differing from the ESP acronym.
Ecosia and Qwant have not disclosed their respective investment figures but have assured that their shareholders are supportive. Importantly, as a separate entity, EUP will not adhere strictly to their previous non-profit business model, enabling the opportunity to attract external capital (if they can entice investors).
The index is anticipated to begin servicing search traffic from France for Ecosia and Qwant by the first quarter of next year, with plans to include a “substantial portion” of German traffic by the end of 2025.
English is set to be the third language implemented, with potential additions of more European languages in the future if demand increases.
Operationally, Qwant’s engineering team will transition to EUP, while Abecassis—who assumed the role of CEO at the search engine just over a year ago—will also lead the joint venture.
Qwant was acquired last year by a cloud technology association named Synfonium, backed by the founders of the French cloud services company OVHcloud, aiming to create a “European champion” in cloud solutions.
In a conversation with TechCrunch regarding EUP, Abecassis clarified that Qwant was already in the process of developing its own search index prior to its acquisition by Synfondium. This initiative will now be transferred to EUP, along with its team and intellectual property assets.
Team-up with Ecosia enhances the likelihood of achieving success, Abecassis asserted, as it broadens the data pool for the index development while also increasing funding and facilitating quicker progress, including additional hiring of engineers.
Ecosia reports having about 20 million users worldwide on a monthly basis, while Qwant caters to around 6 million users in France.
“To truly optimize our efficiency, we need to broaden our reach and become more ambitious,” Abecassis recounted how Qwant approached Ecosia to propose a partnership focused on building the search index.
“For Qwant, this represents a significant opportunity to enhance technology—search technologies flourish when utilized. Increased use drives more investment and generates more data. One of Google’s strengths lies in its vast data volume.”
The two organizations share several characteristics that make their partnership a promising cultural alignment, as both are European alternatives to traditional search engines and their respective business models challenge the data-driven motives of Big Tech. EUP is set to be headquartered in Paris.
“Developing such technology from the ground up is virtually unfeasible,” Abecassis remarked. “The more users and data sets we acquire, the more valuable our technology becomes.”
Kroll emphasized that Ecosia brings expertise, data, and funding to the collaboration—indicating that in addition to the search engine, EUP will need to innovate other technologies, such as widgets integrated into search results.
The partners anticipate that their collaboration will enhance the efficacy of search results delivered to users, while allowing each platform to continue cultivating its unique user experience as EUP refines its ranking algorithms.
Search Ranking Alternatives
Competing search engine Brave, which, similar to Qwant, emphasizes user privacy, has developed its own search index. In fact, it eliminated its API reliance on Bing for text-related searches in April of last year, marketing its service as a genuine alternative to traditional search engines.
When queried about this, Abecassis indicated that Brave’s index closely aligns with Google and Bing’s technical frameworks, while stressing that EUP is being built from the ground up, claiming it will provide more varied search outcomes.
“We do not simply replicate Google or Microsoft’s models,” he asserted. “We thoroughly index all accessible documents, comprehend their content, and have a team dedicated to identifying the optimal connection between a document and a search query.”
“Though shortcuts exist to replicate established technologies, we have chosen a unique path—developing everything from the ground up. While this approach is more challenging, we believe it is ultimately more sustainable.”
A notable difference between EUP and Big Tech search is that EUP’s search index will prioritize “privacy-first” outcomes. What does this practically entail? Abecassis explained that, unlike Google, their technology refrains from personalizing search results based on individual user data.
“We will continue to operate without [personalizing results based on user] data,” he stated. “Our algorithms will then be refined based on the available data.”
“I believe this is a substantial victory—a significant privacy win,” Kroll remarked regarding this technical approach. He also underscored the strategic advantage of having a search infrastructure established in Europe during this period of growing geopolitical uncertainty.
“From a European standpoint… what does reliance on external search infrastructure mean for the European Union’s autonomy? Considering the recent [U.S.] election outcomes… If the U.S. government decided to discontinue providing search results to Europeans, we would be left with no choice but to resort to traditional phone books.”
“While there is a privacy aspect, there is also the vital issue of data sovereignty,” he added. “I genuinely hope that the United States and Europe will continue to be strong allies. However, the future directions of both regions remain uncertain, making this initiative crucial.”
A Costly Endeavor?
TechCrunch inquired about Brave’s decision to build its own search index. The company replied that prior to this shift, it “constantly faced the risk of Microsoft imposing constraints upon us or cutting off access entirely”—thus, this strategic move aimed at alleviating dependency risks.
“Our quality assessment team, which conducts blinded quality evaluations, reports that our results are on par with Google and superior to Bing in the countries we monitor (where Brave Search is the default for Brave browser users),” Brave disclosed, adding that Brave Search is “the fastest-growing search engine following Bing,” processing over 1 billion queries monthly.
On the topic of index development costs, Brave described the process as “prolonged and expensive”—pointing out its acquisition of the open-source Tailcat search engine in 2021, which they indicated has a development history that traces back to 2014.
“It explains why only three fully independent search indexes exist in the Western world,” Brave added.
The company licenses its search index through the Brave Search API, which is reportedly utilized by “numerous leading companies in the AI sector” and is quickly emerging as a “significant” income source.
TechCrunch also consulted with search engineer Peter Popov, who has 15 years of experience at the Russian search giant Yandex, to gauge the costs associated with building a search index. Currently, he serves as VP of ads at VK.
“In very general terms, constructing a search index, factoring in hardware and development expenses, is unlikely to exceed $10 million,” Popov indicated, viewing such an investment as “not particularly large.” He suggested that advancements in AI have simplified the creation of quality search results without relying on vast user-generated data, thanks to modern LLM models equipped with inherent knowledge of search semantics.
However, he cautioned about the growing challenges regarding where search bots can legally crawl, highlighting that a comprehensive search index necessitates broad access to information sources to effectively serve users’ inquiries.
“Proprietary platforms often pose considerable obstacles for gathering information,” Popov remarked to TechCrunch.
“Technically speaking, establishing a search index capable of serving the entire web isn’t overly complex. The expansion of accessible information lags behind advances in computing capacity. One of the challenges around scaling AI stems from this relatively small volume of information.”
“The sphere of useful information on the internet is rather limited,” he continued. “Currently, there’s no extensive directory of websites available for searching. Notably, of these mainstream web platforms, Wikipedia remains one of the few open to search queries.
“Following Wikipedia, there are limited options for useful sites like arxiv.org or substantial online libraries. Such information can serve dual purposes—informing network training or supplying data to the neural network during inference, where search functions as a component of the LLM backend.”
In essence, to ensure a search index is useful, it must be able to crawl the web freely. However, as Big Tech becomes increasingly protective of information within their own ecosystems, this complicates efforts to emerge from their shadows by indexing data from the internet for search functionalities—creating a challenging scenario.
Compiled by Techarena.au.
Fanpage: TechArena.au
Watch more about AI – Artificial Intelligence
