TheChek News Flow

Harvey, a startup building what it describes as an AI-powered “copilot” for lawyers, has raised $100 million in a Series C round led by GV, Google’s corporate venture arm.

The tranche, which also had participation from heavy-hitting angels and VCs OpenAI, Kleiner Perkins, Sequoia Capital, Elad Gil and SV Angel, brings Harvey’s total raised to $206 million and values the company at $1.5 billion.

In a post on Harvey’s official blog, co-founders Winston Weinberg and Gabriel Pereyra said that the bulk of the new capital will be put toward collecting and curating data to build and train “domain-specific” AI models while growing headcount and expanding into new geographies.

“This investment will enable Harvey to continue scaling and improving our AI-powered technology across business functions and geographies,” Weinberg and Pereyra said. “We will use this new capital to invest in the engineering, data and domain expertise that are fundamental to building AI-native systems that facilitate the most complex knowledge work. We will also deepen our partnerships with both cloud and model providers to integrate additional models into Harvey and broaden our training collaborations to continue improving model efficacy.”

Weinberg, a former securities and antitrust litigator at law firm O’Melveny & Myers, and Pereyra, previously a research scientist at DeepMind, Google Brain (another of Google’s AI groups) and Meta AI, launched San Francisco-based Harvey in 2022. Weinberg and Pereyra are roommates; Pereyra showed Weinberg OpenAI’s GPT-3 text-generating system and Weinberg realized that it could be used to improve legal workflows.

Harvey, powered by OpenAI’s GPT-4 model family, can answer legal questions phrased in natural language, like “Tell me what the differences are between an employee and independent contractor in the Fourth Circuit” and “Tell me if this clause in a lease is in violation of California law, and if so, rewrite it so it is no longer in violation.”

Harvey also offers tools that can automatically extract information from trial transcripts, automatically find legal documents that can back up court arguments and generate first drafts of filings that incorporate info and citations from legal databases.

It’s powerful stuff in theory. But it’s also fraught. Given the sensitive nature of most legal disputes, some lawyers and law firms might be reluctant to give a tool like Harvey access to any case documents. There’s also the matter of language models’ proclivity to spout toxicity and made-up facts, which would be particularly poorly received — if not perjurious — in the court of law.

That’s why Harvey has a disclaimer attached to it: The tool isn’t meant to provide legal advice to nonlawyers and should be used under the supervision of licensed attorneys.

Harvey faces some competition. Casetext uses AI, primarily OpenAI models, to find legal cases and assist with general legal research tasks and brief drafting. More surgical tools like Klarity use AI to strip drudgery from contract review. At one point in time, startup Augrented was even exploring ways to leverage OpenAI models to summarize legal notices or other sources in plain English to help tenants defend their rights.

But Weinberg and Pereyra claim that Harvey is flying high, in use “daily” by tens of thousands of lawyers at law firms and consultancies including Allen & Overy, Macfarlanes, Ashurst, CMS, Reed Smith and PwC. Annual recurring revenue has tripled since last December, the two co-founders said in the blog post, while Harvey’s workforce has tripled in size.

The Information reported in early June that Harvey hoped to raise $600 million at “at least” a $2 billion valuation, in part to acquire a legal research service called vLex to train its AI products. Those plans fell through — hence the vastly reduced Series C.

On Monday, Cohere, a Canadian large language model developer, announced it had raised $500 million in a deal led by Canadian pension fund PSP Investments at a $5.5 billion valuation, confirming our report from March.

That valuation isn’t cheap. Around the time of the investment, the startup was generating $35 million in annualized revenue, according to a person briefed on the financials. (Annualized revenue typically reflects last month’s revenue multiplied by 12, implying it was generating about $2.9 million per month as of March.)

That means the pension fund valued Cohere at more than 140 times its forward revenue—that's a lot even by AI standards. For instance, OpenAI and Anthropic—both of which generate many times more revenue than Cohere—were last valued at 54 and 75 times their forward revenue, respectively. (Check out this past newsletter for more on how investors have been valuing AI startups.)

Are the Canadian investors off base? It’s not that straightforward.

PSP Investments has backed a handful of established companies, such as security firm Netskope and ride-sharing app Lyft. But, it’s not exactly known for its early-stage startup investing.

However, the investment looks more reasonable when you take into account the intangible benefits PSP Investments—and Canada more broadly—gets from this deal.

By investing in a homegrown model developer, Canada is ensuring that it can forge its own path in AI development, separate from U.S.-centric AI startups and regulatory environments in the States and the EU that might disadvantage foreign AI companies.

We could see signs of that months ago, when the Canadian government announced in April $2 billion-plus in subsidies (including computing power), to benefit local AI startups, including Cohere, according to a person with direct knowledge of the subsidy allocations.

So at a minimum, PSP Investments is lifting up a local company that perhaps will help Canada have a voice in the AI race. In the worst case scenario, if Cohere doesn’t work out, whatever the fund loses would be a minuscule percentage of the $265 billion in assets it manages. Seems like a risk worth taking!

Perhaps the people with the most to lose are new and future employees (who will receive stock tied to this valuation) and Cohere’s newest investors that don’t have ties to the Canadian government, such as Strategxy Ventures or IronArc Ventures. These parties are hoping to cash out through some kind of sale or IPO in the future. To make money on their shares, they’ll need Cohere’s already-steep valuation to double or triple so they can make a reasonable return for their own investors.

Getting to that goal may test their patience.

Here’s what else is going on…

Perplexity Perplexes Media Outlets

Yesterday, Sahil and I broke the news that Condé Nast, owner of The New Yorker, Vogue and Wired, has sent a cease-and-desist letter to AI-powered search engine Perplexity, demanding that it stop using content from Condé Nast publications in its search results. It follows a similar letter from Forbes accusing the startup of infringing its copyright.

However, the questions about Perplexity’s data sources go beyond media outlets and publications. As we earlier reported, others have criticized Perplexity for claiming to take on Google while using its ranking signals to order its own search results.

It calls into question how much AI startups are depending on copyrighted data and technology developed by their suppliers and rivals—and, what that means for their long-term survival.

Regulator fined macro hedge fund £41mn following allegations of ‘reckless’ conduct

The UK’s financial conduct regulator is fighting to resurrect a $700mn redress scheme that it wants to impose on billionaire Michael Platt’s BlueCrest Capital over allegations of “reckless” conduct.

Three senior judges in London on Tuesday heard a legal challenge brought by the Financial Conduct Authority after a tribunal last year found the regulator had demonstrated “a considerable amount of muddled thinking” and blocked the regulator’s redress plan.

The FCA fined BlueCrest, once one of the world’s best known macro hedge funds, £41mn and proposed the client-redress scheme in 2021 over an alleged conflict of interest.

The regulator said the firm failed to manage the risk that moving managers of a fund for external investors, to a fund that invested money on behalf of partners and employees, could disadvantage the outside clients.

BlueCrest, which denies the findings, launched a legal challenge. The tax and chancery chamber ruled last year that the regulator lacked powers to impose the redress scheme.

The two tribunal judges, Timothy Herrington and Rupert Jones, criticised the regulator for a “lack of clarity” that made it “difficult to identify the essence of the authority’s thinking”. They called the FCA’s decision notice “not an impressive document”.

The FCA is in turn challenging the tribunal’s decision, and Lord Justice Popplewell, Lord Justice Nugee and Lady Justice Falk in the Court of Appeal heard the case on Tuesday.

Andrew George KC, representing the regulator, told the appeals court that the tribunal judges had made errors in reaching their decision.

George said the legislation “clearly and expressly” grants the FCA powers to secure an “appropriate” degree of protection for consumers. The tribunal, he argued, had wrongly placed “complex and unnecessary” constraints on the regulator’s powers.

However, Javan Herberg KC, representing BlueCrest, said the tribunal’s decision to block the $700mn redress scheme had been correct, and called on the court to dismiss the FCA’s appeal.

Herberg said the regulator’s penalty had been “disproportionate and excessive” as the regulator had “not sought to assess the loss actually caused” nor “shown that any loss was caused by the firm’s wrongdoing”.

The FCA, in its original decision, claimed BlueCrest’s disclosures to investors had been “entirely insufficient and, at times, misleading”.

Decisions about allocating traders to the internal fund were made by senior managers who also invested in that fund, which “placed them in a situation where they stood to benefit” personally, the FCA said at the time.

The regulator found that a BlueCrest algorithm called Rates Management Trading, designed to replicate trades made by the firm’s traders working on the internal fund for the external fund, at times underperformed.

The regulator, whose decision covered the period from October 1 2011 to December 31 2015, described the firm’s conduct as “reckless rather than deliberate”.

BlueCrest, which at its peak ran about $36bn in assets, announced at the start of December 2015 that it would stop managing money for outside clients.

The firm, co-founded by Platt, 56, converted to a family office following a drop in assets and a period of weak returns.

The Italian group’s scheme of setting up satellite businesses is starting to show promise

How to traverse the energy transition is the multibillion-dollar question that all of Europe’s oil majors have been grappling with for years. Nobody has cracked it — as the recent strategic revisions from the likes of Shell prove.

But one strategy is starting to show promise: Italian major Eni has taken the approach of setting up satellite businesses, which can independently tap capital markets and operate unencumbered by the rest of the legacy business.

Eni is in talks with KKR to sell a 20 to 25 per cent stake in its biofuels business Enilive. The deal, if agreed, could potentially value the entire satellite, which also owns more than 5,000 fuel stations in Europe, plus cafés and a car-sharing business, at between €11.5bn and €12.5bn, Eni said.

Debt-free, Enilive is expected to generate €1bn in ebitda this year. This would imply a forward enterprise value/ebitda multiple of 11.5 to 12.5 times, well above other biofuels businesses such as Neste, which currently trades on about 7 times.

Granted, pure biofuels companies have been hit in the last year thanks to an oversupplied market. Enilive stands out as it produces some of its own feedstock, points out Bernstein’s Irene Himona.

But even when taking into account a higher multiple for Enilive’s other operations, a deal at that level would be a good outcome. Eni last year also sold a stake in its renewables-focused satellite Plenitude, also at an unexpectedly attractive multiple.

Those should be two wins in a sector that is struggling to convince traditional fossil fuel investors of the value to be had in clean energy businesses. But the market reaction to the KKR talks was almost zero. Eni — at some point — deserves more credit.

In the past year, the Italian group’s shares have underperformed several rivals, including TotalEnergies and Shell. Other factors have been at play, including the hit of weaker gas prices to earnings estimates and negative surprises on net debt, notes UBS’s Joshua Stone.

Oil investors’ preoccupation with just securing more share buybacks also plays its part — despite Eni recently improving its distributions policy to 30 to 35 per cent of cash flow from operations, versus 25 to 30 per cent previously.

Eventually, Eni is likely to float its satellite units. Equity market sentiment towards clean energy businesses clearly is not good enough to support that as yet. For now, it is doing well at making the most of the plentiful private capital on offer to show that, ultimately, there is value there to be had.

Isaac “Ike” Perlmutter, the former Marvel Entertainment executive, who for the last decade and a half has been one of Disney’s largest independent shareholders, has sold his entire stake.

Perlmutter banded with activist investor Nelson Peltz last year in a campaign to secure two seats on Disney’s board of directors, contributing nearly 26 million shares to Peltz’s war chest. In April, shareholders voted overwhelmingly in favor of Disney’s slate, effectively endorsing CEO Bob Iger and his management team while handing Peltz and Perlmutter a bruising defeat.

Perlmutter sold his entire position of 25.6 million shares in the months after the vote, according to communications with his investment adviser seen by The Wall Street Journal.

He became a major shareholder in the entertainment giant in 2009 when he sold Marvel Entertainment to Disney for more than $4 billion. At the time, Disney shares traded at $26.84. Since then, his company stock has appreciated by more than $2 billion. Perlmutter said he sold his stake between early April and mid-July at an average price of just under $115.

Perlmutter said in an interview that he sold because he doesn’t have confidence in Disney’s current management and expects the company’s share price and financial performance to decline further. If Disney’s shares decline to $65 to $75 per share, he plans to buy much of his old stake back.

Disney shares were down 3.8% in Tuesday morning trading. They are down 26% since early April, when they closed at a 52-week high of $122.82 just before the shareholder vote. Disney declined to comment.

The digital banking app provider has begun to churn out profits thanks to user growth and rising interest rates

Bank disrupter Revolut is nearing a deal to sell about $500 million of employee-owned shares at a valuation of $45 billion, according to people familiar with the matter, a sign that animal spirits are returning to the financial-technology sector.

The details

Revolut, already the world’s second most-valuable fintech startup after Stripe, is in talks to sell the shares to new shareholders, including New York-based Coatue Management, and existing investors such as Tiger Global Management.

Revolut, which competes with incumbent banks for retail and business customers, has also been in talks with investment firm Greenoaks as a potential participant. A deal, if completed, could be announced in the coming days and would pave the way for a potential initial public offering.

The stock sale would increase Revolut’s valuation by more than a third. The company was valued at $33 billion in 2021 when it raised $800 million from a group led by Tiger and SoftBank Group’s Vision Fund unit. Japan’s SoftBank isn’t buying shares in the latest offering.

The rationale

Revolut won’t gain fresh funds from the share sale to plow back into its operations. Instead, the effort allows long-term employees to cash out some of their holdings, while setting a floor for the company’s potential valuation ahead of any IPO.

Startup employees are often compensated heavily in shares. Typically, staffers would wait for the company to go public to cash in. The global IPO market, though, has struggled in recent years because of high interest rates and the flops of some high-profile new listings.

The context

Documents provided to investors show that Revolut has begun to churn out profits. It reported $429 million of net income in 2023, and forecasts $976 million of profit in 2024.

Revenue grew to $2.2 billion in 2023 and is projected to hit $3.7 billion in 2024, as businesses and households increasingly migrate to its digital-only finance platform.

Founded in 2015, U.K.-based Revolut operates a banking-like app that lets customers deposit money, make payments, and buy and sell currencies, stocks and crypto. The company, led by co-founder Nik Storonsky, has a European Union banking license, but has struggled to secure one for its largest market, the U.K. It stumbled with its accounting practices, which it has since moved to strengthen to win regulator approval.

The company has been aided by higher interest rates—19% of its revenue came from interest income in the 12 months through March—and a growing roster of users around the world. It reported over 20 million customers actively using the app as of March, up from 9.2 million in early 2022.

A $45 billion valuation is lofty measured against some of its rivals. It is about 17 times the company’s revenue over the 12 months through March. That compares with just over two times for PayPal and almost eight times for Nu Holdings, a Brazilian-based digital bank listed on the New York Stock Exchange.

Revolut’s goals are even loftier. It projects $9.3 billion in annual revenue in 2026 and has a long-term ambition of hitting $100 billion in sales by 2040—which would be over three times as large as PayPal’s 2023 revenue.

The investment comes despite broader struggles in the fintech sector, where valuations have tumbled since interest rates began rising, while profits have proved more elusive than once hoped.

National Audit Office report says strategy does not deliver value for money

The UK’s spending watchdog has questioned whether Britain’s investment in the European Space Agency is delivering value for money, after noting that the country received a lower return than France, Germany and Italy over most of the last decade.

The National Audit Office on Tuesday published a highly critical assessment of progress on implementing the UK’s national space strategy, unveiled in 2021 by the previous Conservative government. While directing space spending through the ESA “gives UK space companies and academics access to large-scale space programmes which would be challenging to replicate nationally,” the report said, “the UK does not yet receive contracts . . . proportionate to the value of the funding [it] provides.”

In addition to questioning the returns from ESA membership, the NAO said a lack of clarity and detail in the national space strategy was undermining the UK’s ambition to expand its space economy. Three years after unveiling the strategy, the UK Space Agency and the Department for Science, Innovation and Technology were still in the early stages of “identifying and developing the plans and capabilities needed to deliver the strategy’s ambitions”.

The NAO’s findings will rekindle a debate about the value of ESA membership after British companies were excluded from bidding for contracts on navigation service Galileo and the Copernicus earth observation project as a result of Brexit. 

Although UK companies are again eligible for Copernicus contracts since January, many in the space sector have been lobbying for a strong national programme to accelerate growth in areas where the UK has expertise, such as earth observation and data analytics.

“We still do not have a big national project that industry can focus on,” said one veteran space industry executive.

ESA acts as the procurement agency for the EU, but is an independent intergovernmental organisation that pools investment in big scientific and public space programmes. Its 22 members include non-EU nations such as the UK, Canada and Norway. 

Britain is the fourth largest contributor to ESA, accounting for about 9 per cent of its non-EU funding. Roughly three-quarters of the UKSA’s budget goes to ESA programmes and in 2022 the previous government pledged £1.84bn in funding between 2023 and 2027.

The ESA aims for member states to receive the equivalent of their funding pledges in contracts to industry and academia, minus a small deduction for overheads. A confidential report by the UKSA’s executive committee found that between the final quarter of 2015 and February 2023 the UK received 93p back in contracts for every £1 allocated to the ESA. 

The NAO notes that the UKSA has made efforts to boost the returns, helping companies to identify opportunities and to submit applications. By the end of last year the return had improved to 96p on the pound, “but this means UK industry and academia are still not benefiting in terms of contract value from the full funding given to ESA by UKSA,” the NAO said. 

“France, Germany and Italy achieved values between 0.99 and 1.02,” the report added.

David Parker, ESA’s Director of Human and Robotic Exploration until 2023, told the Financial Times that the UK received substantially more economic value from its ESA membership than was accounted for by contract awards. “The UK has had a larger share of the return on science programmes,” Parker said.

Parker said some of the deficit reflected a “change in the UK industrial landscape. Some of the capabilities have changed.” But work was being done to educate companies about the opportunities and procedures. “It will come back again,” he said.

The UKSA said it was working to improve the returns, with a target to reach parity by the end of this year. “We are making good progress,” the agency said.

Josef Aschbacher, director-general of the ESA, told an audience at the Farnborough air show that the agency was taking steps to “give more energy” to its UK based centre focused on satellite applications and telecommunications.