An exclusive Q&A with AI upstart and OpenAI competitor Cohere’s CEO, and more tech and AI news this week

Welcome back to LinkedIn News Tech Stack, which brings you news, insights and trends involving the founders, investors and companies on the cutting edge of technology, by Tech Editor Tanya Dua. You can check out our previous editions here.

Pitch me the interesting investors, founders, ideas and companies powering emerging technologies like AI to reach the inboxes of nearly 1 million subscribers plus thousands more on LinkedIn. Follow me for other tech updates. And click 'Subscribe' to be notified of future editions.

ICYMI: Catch this week's edition of VC Wednesdays with New Enterprise Associates (NEA)'s Aaron Jacobson here.

Publishing Note: Tech Stack will be on pause for the July 4th holiday next week, and will resume in July. Click 'Subscribe' to be notified of future editions.

A deep dive into one big theme or news story every week.

Cohere may not be a household name like OpenAI’s ChatGPT, but that doesn’t concern co-founder and CEO Aidan Gomez. 

That’s because the Toronto-based startup, which also builds large language models competing with the likes of Google's Gemini, Meta’s Llama family of models and Anthropic’s Claude, is going squarely after enterprises — focusing on the business applications of AI.

Armed with $450 million in fresh funding (which he declined to comment on), the upstart is looking to scale its sales operations at a time when more businesses are putting AI initiatives into production, Gomez told me in a sit-down interview at the Collision Conference in Toronto last week. 

Catch the full conversation below.

Tanya Dua: Do you consider yourself to be an underdog in the AI race?

Aidan Gomez: Absolutely. Cohere is, relatively speaking, still under the radar. We just focus on doing the work the world needs to adopt the technology [AI] — building the platform that the enterprise market needs in order to succeed. We do that work quietly and don’t really make a fuss about it. The tech is coming to fruition now; it's finally getting out of this pilot phase and into production. And that's genuinely happening across hundreds of thousands of organizations, and impacting millions.

TD: Why is Cohere the right partner for those organizations?

AG: We're focused on the problems the enterprise cares about. The use cases inside the consumer world and the enterprise world are very different. At the technological layer, the data itself is different. For instance, we work with an HR software platform to power a tool that helps them draft employment contracts faster. That’s not a use case consumers care about, who are looking for the news of the day or help with their homework. We exclusively focus on use cases that enterprises care about, and that means our technology outperforms competitors who are focused on consumer use cases.

TD: Why the enterprise focus? Is it because that’s where the money is?

AG: We’re a business, it’s not about wanting to make money, we have to make money. But What Nick Frosst, Ivan Zhang and I [Cohere's co-founders] wanted to do was build the platform to push the technology out there. We were really frustrated with the lack of attention and adoption it was getting from the world, and this was pre-ChatGPT and shortly after the Transformers paper came out. Our prerogative was to get AI out there the furthest, the fastest and in the most impactful way we could — and that was by helping everyone else adopt it. Our pitch to large enterprises is that if you don't succeed or if you don't push this proof of concept to production, we don’t succeed. So, we're going to do whatever it takes for you to succeed. 

TD: What would you say is the biggest challenge to enterprise adoption of AI, and how is Cohere stepping up to meet that?

AG: Trustworthiness – getting enterprises to get comfortable that the model is going to be reliable. That you can trust what it says. And if it makes a mistake, it'll do that in a transparent way and not in an opaque way. It needs to be able to cite its sources. And so, we've invested super heavily in RAG, or retrieval-augmented generation [a popular method of optimizing the output of an LLM that makes it reference an authoritative knowledge base outside of its training data sources before generating a response]. Patrick Lewis, who leads our RAG team, created it, and it's crucial to trust because it's the way that you can audit a model. So when it says something, it’s forced to justify why it's saying that and humans can check its work.

TD: You recently rolled out a program geared toward startups. Is that to expand your customer base?

AG: Startups are always the first adopters. Before any enterprises even woke up to AI, startups were already building, experimenting and creating and deploying new experiences. So, they're the most exciting party to work with because they're so invested, and it’s so high stakes for them for the technology work. We want to support them, make this tech cheap for them to adopt and want to make sure that they succeed with their use cases and products. But we can’t do that for hundreds of them. What we can do is create a program, pick and choose and build an accelerator so that we can deeply invest in a group of them.

TD: Are you planning to scale your go-to-market motion further with the recent funding?

AG: Growth has been very, very exciting to see. Enterprises are adopting this technology. Spending is going up. The workloads are shifting from the proof of concept to actual production, and we're doing a good job at winning those. So we’re definitely scaling our go-to-market motion at the moment. Right now, we have shockingly few salespeople, so we really need to grow that. But we have fantastic leadership. Meini Heuberger leads that team. He led Cisco’s sales operations, which is a multibillion-dollar business, so he knows how to scale the team to many thousands of salespeople. But right now, we’re talking about scaling the team to double digits.

TD: The Canadian government recently announced an investment of nearly $2 billion in the country's AI sector — something that Cohere will benefit from. What’s your view on the concept of sovereign AI, and the idea that controlling digital technology is critical to defending and promoting national interest?

AG: From the perspective of the government itself, I think it's super important, especially if your citizenry doesn't speak English, right? Because if you believe this technology is going to augment your workforce and if it doesn't speak the language of your workforce, you just miss out on a productivity multiplier. Your enterprise competitiveness, global competitiveness and worker satisfaction drops. So it should be a national imperative. It makes a lot of sense.

Here’s where we bring you up to speed with the latest advancements from the world of AI.

• Amazon is reportedly building a rival to ChatGPT, Business Insider reports, citing anonymous sources. The project, code-named Metis, uses the retrieval-augmented generation method to create a bot that can provide more up-to-date responses to queries from customers. “Metis should be able to share the latest stock prices,” BI wrote. But the e-commerce giant has struggled in the AI race so far, with mixed reviews for its corporate chatbot and performance issues around its AI chips, BI has reported. The tech giant plans to offer the customizable bots as soon as this year. Meanwhile, Google is building a product that lets consumers create their own chatbots, choosing personality and physical traits — similar to an offering by startup Character.AI.

• Speaking of ChatGPT, OpenAI is pushing back the launch of its already controversial voice assistant, saying it needs to do more safety testing. The company originally planned to offer the feature to some paying ChatGPT subscribers at the end of June, but will now wait another month. All paying users will be able to access the assistant in the fall. OpenAI demoed the feature last month, but quickly came under fire when critics claimed one of its synthetic voices sounded too much like an AI bot played by actress Scarlett Johansson in the 2013 movie "Her." Relatedly, OpenAI seems to be on a buying spree, and has nabbed collaboration platform Multi days after acquiring database tech startup Rockset. Multi lets users share screens and work jointly on code, and its acquisition could signal an intent to build business-focused real-time collaborative features, The Information reports. Meanwhile, OpenAI's earlier acquisition of Rockset is expected to beef up its data processing, analytics and retrieval infrastructure, speeding up its data ingestion and analysis capabilities as it trains new AI models. 

• Major record labels including Universal Music Group, Warner Music Group and Sony Music Entertainment are suing two AI startups for copyright infringement, Bloomberg News reports. The Recording Industry Association of America filed the lawsuits on behalf of the labels, accusing Suno and Udio of using copyrighted music to train their AI models without permission or compensating the artists. The RIAA is seeking as much as $150,000 per infringed song, an amount that could potentially reach billions. Suno, which also has a partnership with LinkedIn’s parent company, Microsoft, lets users leverage generative AI tools to automate music-making.
• NVIDIA seesaws in the market. Nvidia’s shares rebounded from a three-day downtrend that saw the chip maker's market value plunge by $430 billion. The stock advanced 7% on Tuesday, leading the technology sector higher. The recent selloff in Nvidia, which had sent its shares into a technical correction, sparked concerns over the company’s influence on the S&P 500, having driven a third of the index's 2024 gains. While the stock is still up significantly on the year, analysts have warned that a sustained Nvidia selloff could trigger a broader market downturn. 
• xAI and Dell Technologies are planning a 'supercomputer.’ Shares of Dell and SuperMicro got a boost last week after Elon Musk announced the two hardware giants will team up with his AI startup xAI to develop a supercomputer. Dell will provide half of the server racks for the project, Musk said. xAI is planning a $500 million supercomputer in Buffalo, N.Y., and a "supercomputer cluster" in Austin, Texas, that could help Tesla create its long-anticipated autonomous vehicles. Separately, Dell CEO Michael Dell revealed that his company will build an "AI factory" with Nvidia to power xAI's chatbot, Grok.
• Peacock will offer personalized daily recaps of the Paris Olympics narrated in the voice of Al Michaels, generated by AI. The legendary sports announcer gave his blessing to the endeavor, marveling that the AI version of him was nearly perfect, if "a little bit frightening," according to Vanity Fair. An in-house team built the feature, training the AI voice on Michaels' NBC appearances, and editors "will review all content," NBCUniversal says.

Here’s a list of other notable AI developments from this week:  

• AI chip startup Etched, which aims to take on Nvidia with its specialized chip geared toward transformer models, has raised $120 million from investors including Primary Venture Partners and Positive Sum Ventures to expand manufacturing, the company announced on Tuesday.
• A new AI biotech startup called EvolutionaryScale founded by former Meta researchers has emerged from stealth, raising $142M in seed funding to help scientists craft new molecules. See investor Patrick Hsu’s LinkedIn post on the announcement here.
• AI-powered recruitment firm TechWolf has raised $43 million in Series B funding, led by Felix Capital.
• AI compliance startup Norm Ai has raised $27 million in series A funding by investors including Coatue and Bain Capital Ventures. See CEO and founder John Nay’s LinkedIn post and investor Rak Garg’s LinkedIn post for more details.
• Meta has launched a new AI accelerator program to help European startups integrate open foundation models like Meta Llama into their products. See Meta’s VP of GenAI Ahmad Al-Dahle’s LinkedIn post here.

Catch up on the tech headlines you may have missed this week and what our members are saying about them on LinkedIn.

• The EU is taking aim at Apple under a new law. Apple is accused of preventing developers that make games and other applications for its App Store from steering consumers to cheaper services, in violation of a new European Union law known as the Digital Markets Act that was passed in 2022 and targets Big Tech companies. If found to be in breach, Apple would be the first company charged with violating the law, and could face a fine of up to 10% of global revenue, or up to 20% for repeat infringements. The European Commission also said it started a new probe into Apple's contractual terms with developers. Apple says it is "confident our plan complies with the law." The EU will make a final decision by March 25. Late last week, Apple said it would not release new AI features in the region due to regulatory uncertainty. The tech giant has also been talking with Meta about integrating its generative AI model into Apple Intelligence, The Wall Street Journal reports, citing an anonymous source.
• Reddit, Inc. is cracking down on bots that scrape its data to train AI models, The Verge reports. The platform plans to enforce stricter compliance to its existing policy for how commercial enterprises access its public data, requiring a licensing agreement like those held by OpenAI and Google. Websites typically have what’s known as a robots.txt file, which allows or blocks search engines and automated bots from accessing data; Reddit’s will now block unlicensed crawling. The move comes as AI startups, seeking to train their models, are increasingly under fire for ignoring this established framework. 
• TikTok's biggest advertisers are scrambling to create backup plans in case the app is banned in the U.S., the Financial Times reports. TikTok generated $16 billion in U.S. sales last year, mainly from advertising, anonymous sources tell the paper. For many brands, the app is a crucial part of their marketing efforts because it has such a huge reach — more than 1 billion users worldwide — and helps them connect with younger audiences. Advertising agencies are trying to game out where TikTok users might migrate after a ban, such as Instagram or YouTube. It's not just advertisers who might suffer: Oracle is warning investors that a TikTok ban would hurt its bottom line, as TikTok is thought to be one of its largest cloud customers.
• SHEIN has confidentially filed for a London listing in what could be one of the world's biggest initial public offerings of 2024, Reuters reports, citing anonymous sources. The Chinese-founded fast fashion retailer, valued at $66 billion last year, set up the paperwork in early June. It's not yet known when Shein plans to launch the IPO, but it is expected to be the biggest the U.K. has seen in about a decade, and comes after the company deemed it unlikely that the Securities and Exchange Commission would approve a New York IPO, per Bloomberg. Shein confidentially filed for a U.S. offering late last year, but has since faced scrutiny over the use of forced labor in its supply chain and its abuse of a U.S. tax law, notes CNBC. Speaking of Shein, Amazon is planning to launch a new section featuring cheap, unbranded items shipped from China, The Information reports, citing anonymous sources. The move is Amazon's "most aggressive response yet" to popular competitors like Temu and Shein.  Amazon plans to onboard Chinese sellers this summer, with shipping times of nine to 11 days. Meanwhile, Amazon's valuation topped $2 trillion for the first time Wednesday — a feat that took four years to cross — as shares surged almost 4%.
• Netflix is rebooting its famous ‘culture memo.’ The latest version of the company’s 2009 declaration of the company's values and ethos was vetted for eight months and received 1,500 comments from employees, according to Netflix's talent chief, Sergio Ezama. The new memo sticks to the company's philosophy of "People Over Process," but with some concessions to its new reality as a 13,000-person workforce rather than a scrappy startup. For example: The newly added declaration that "not all opinions are created equal," because it's no longer realistic for every employee to weigh in on every decision. Netflix co-founder and former CEO Reed Hastings explained: "We care about freedom when it generates excellence, not for its own sake. In hindsight, this is the draft I wish we had 15 years ago."
• Barcodes are getting a new, 50-year upgrade. It started with a beep half a century ago when a pack of chewing gum was scanned in an Ohio supermarket. Since then, barcodes have become an everyday part of our lives, and now, they’re getting an upgrade. GS1 US, which oversees UPC barcodes, is changing the familiar look of numbers and straight lines to 2D barcodes. The 2D versions are similar to QR codes but contain more information, such as details about a product’s ingredients. The transition to the 2D barcode system is already underway, but GS1 US expects the rollout will be complete by 2027.

Here’s keeping tabs on key executives on the move and other big pivots in the tech industry. Please send me personnel moves within emerging tech.

• NVIDIA ’s head of investor relations and strategic finance Simona J. [Jankowski] is leaving the company after seven years, she announced in a post on LinkedIn. 
• General Motors’s Cruise has named former Amazon, Microsoft and Sonos executive Marc Whitten as its new CEO. See Whitten’s Linkedin post here.
• HP has named former JPMorganChase and Medtronic executive Karen Parkhill as its CFO. See HP president and CEO Enrique Lores’ LinkedIn post here.
• Microsoft tech and AI executive Jean-Philippe Courtois, its EVP and president of national transformation partnerships, is leaving the company. See his LinkedIn post here.
• Duolingo has added Mario Schlosser, the co-founder and CTO of Oscar Health, to its board. See Duolingo CTO Severin Hacker’s LinkedIn post here.

As always, thanks for reading. Please share Tech Stack if you like it! And if you have any news tips, find me on InMail.