Dry Powder?
A reminder for new readers. That Was The Week collects the best writing on critical issues in tech, startups, and venture capital. I select the articles because they are of interest. The selections often include things I disagree with. The articles are only snippets. Click on the headline to go to the original. I express my point of view in the editorial and the weekly video below.
This Week’s Video and Podcast:
Content this week from @kteare, @ajkeen, @bgurley, @jasonlk, @ericnewcomer, @coryweinberg, @KateClarkTweets, @GeneTeare, @hunterwalk, @willknight, @omooretweets, @illscience, @klakhani, @jaygoldberg, @adhutchinson, @jessicakmathews, @alex, @cjgbest, @hamishmckenzie,
Contents
Editorial: Dry Powder?
OpenAI Shoots To No. 1 Spot On Private Cloud Startup Ranking, Underscoring AI’s Sudden Ascendency
Generative AI Is Making Companies Even More Thirsty for Your Data
Apple seeks to bolster expertise in generative AI on mobile devices
AI Won’t Replace Humans — But Humans With AI Will Replace Humans Without AI
Editorial: Dry Powder?
The past two weeks have focused on the assessment of Venture Capital and where it is. The headlines have told the story. Last week's Bonfire of The Unicorns? The week before Never Waste A Good [Venture] Crisis and before that Data-First Venture is Coming, a more optimistic view of the future.
This week we go with Dry Powder. Bill Gurley catalyzed the title with his X post about the belief that Venture Capitalists will deploy the “dry powder” they have under management. Gurley explains why the belief that uninvested fund commitments have to be deployed is incorrect.
The key phrase:
The money isn't actually at the VC firm, they are still sitting in the coffers at the LPs. No VC firm I have ever been exposed to feels "pressure" to "get dollars to work."
and
There are also new market realities. Public comps have changed materially, & founder expectations have not moved as fast. Whole industries trade at a fraction of former multiples. So in many cases there simply isn't a market clearing price. This takes time.
Gurley’s comments were in response to many commentators referring to the $271 billion in un-invested venture fund commitments.
The lack of a market clearing price means that there is no price at which some companies could raise capital. This is not about the need for down rounds but about the fact that even a down round could not get priced.
This is why I think the class of 2023 Seed rounds is the new norm. These rounds did not go through the 2020-2022 price regime. They can go on to do A and B rounds and will do so during the next two years.
Some companies will, of course, be able to do down rounds, and many will be led by existing investors or insider rounds. This week we cover Gené Teare’s (Happy Birthday) article on the collapse of global venture funding in July and Jason Lemkin’s X thread on why entrepreneurs should lead the re-capitalization of their companies and lower the valuations before seeking new capital.
In my reading, I found little in the way of optimism.
That said, this is a time of enormous innovation and value creation. It is inconceivable that substantial investments and outcomes will not emanate from investments made in 2023. The idea that this is a great time to invest is not hubris. But the ability to work closely with the best seed investors and partner with them will be key to finding the companies that will make up this group.
The game has changed to a game where late-stage growth investing is no longer driving momentum. Instead, seed investing, including pre-seed, is the name of the game. This cohort's future Series A and B rounds are where to look for growth and value. Great companies will be built and funded.
Substack is this week’s startup of the week. It is not the first time. The founding team made some announcements of intent this week:
“Readers might be the internet’s most important constituency, and yet they have been treated like paupers. The companies that dominate today’s cultural sphere seem to consider the reader’s mind only insofar as it can be tricked into ceding its bounty of attention—the black gold of the “Information Revolution.” What a disappointing revolution it has been. We thought we were going to live in the Library of Alexandria; instead we’re wading through its sewers hoping to find wisdom on wet pages pasted to the walls. “
These intentions mirror a lot of what I have spoken about here and on the
. Writing requires reader engagement. Substack is right to focus on reader features, including the reader economy (what happens when a reader shares your writing with another reader). I look forward to seeing more.Essays of the Week
Dry Powder is not About to be Invested
Down Rounds Should Be Led by Entrepreneurs
Why Bill Gurley & Josh Wolfe Think VCs Won't Deploy All Their Dry Powder Anytime Soon
Read my text conversation with Lux's Wolfe about venture 'wet powder'
ERIC NEWCOMER, AUG 10, 2023
A running question in this newsletter for the past year has been how much dry powder is there out there, really? Venture capital firms raised humongous funds — but how quickly would they deploy them?
As I reported last month, venture firms have taken very different approaches when it comes to deploying the money they’ve raised. Some firms have basically halted investments, while others are still cutting checks. Almost everyone, though, has slowed down significantly. And to make matters worse, the IPO market — an often under-appreciated source of funding for late-stage startups — has slowed, with capital raised down 36% year-over-year.
Benchmark’s Bill Gurley argued convincingly this week that just because venture capital firms have raised big funds doesn’t mean they feel that much pressure to actually deploy them.
He wrote, “There is no urgency to draw them down. The money isn’t actually at the VC firm, they are still sitting in the coffers at the LPs. No VC firm I have ever been exposed to feels ‘pressure’ to ‘get dollars to work.’”
He continued in a series of tweets:
On the back of a market reset, & w/ portfolio valuations being slashed, GPs are mostly sharing bad news w/ LPs. No GP wants to look aggressive/carefree. Imagine being a teenager with two speeding tickets & a fender-bender insisting on taking the new family car out Saturday night.
Additionally, LPs are in a tough spot from a liquidity perspective. New tax laws & mandates insist they pay out ~5% each year to their constituency. Meanwhile, outbound liquidity from VCs (IPOs/M&A) are at a 15 year low (all but stopped). GPs know this.
There are also new market realities. Public comps have changed materially, & founder expectations have not moved as fast. Whole industries trade at a fraction of former multiples. So in many cases there simply isn't a market clearing price. This takes time.
Lastly, startup cap charts are very flexible when prices are rising, but quite brittle/problematic when prices fall. This is due mostly to liquidation preference. Most investors will simply “pass” vs stepping into this complexity.
Anyway, I wouldn’t expect a massive rebound due to this perceived “dry powder.” VC crashes seem instantaneous. Rebuilds take time as the industry slowly works through previous sins, & slowly regains confidence. Risk off is very fast. Risk-on is very slow.
In short, venture capitalists are under no obligation to invest the money that they’ve raised and limited partners (the people who fund venture capital firms) have plenty of reasons to want VCs to invest less.
Lux Capital’s Josh Wolfe has talked about the term “wet powder.” He told TechCrunch, “I say it is ‘wet’ because if 2023 resembles the dot-com bust, all the money raised at valuation peaks will get spent propping up ‘walking dead’ zombie companies that can’t raise outside money and instead have to turn to inside rounds.”
‘Tidal Wave’ of Down Rounds Hits Startups
With cash running low, founders are conceding to sharply lowered valuations. The moves can slash their prices by as much as 96%.
By Cory Weinberg and Kate Clark
Aug. 4, 2023 6:01 AM PDT
Turntide Technologies, a maker of electric motor systems backed by Bill Gates–founded Breakthrough Energy Ventures, was one of the more than 300 private companies that passed the $1 billion valuation threshold last year, putting it firmly into unicorn status. It’s not worth close to that anymore. The Sunnyvale, Calif., firm is nearing a deal for new capital that would likely slash its valuation by more than 80%, according to corporate filings and executives.
The trickle of down rounds that has hit the startup world over the past year is starting to become a flood, investors and lawyers say. In the most extreme cases, companies like Turntide are planning “cram-down” rounds of financing, which heavily dilute some existing investors’ stakes with diminished valuations. Investors at autonomous-delivery startup Nuro have discussed a similar move. Likewise, internet answer site Quora has recently wiped away some of its investors’ preferred ownership rights in a corporate filing, a move that often foreshadows cram-downs, lawyers say.
THE TAKEAWAY
• Quora files paperwork that often signals cram-down
• Fintech Rho quietly cuts valuation about 10%
• Nuro investors have discussed major down round
“We’re about to see a tidal wave of these companies doing reset rounds,” said Greg Smithies, a San Francisco Bay Area–based partner at Fifth Wall, an investor in Turntide Technologies and other climate and property tech startups. “It’s just going to get worse before it gets better.”
More executives are overcoming the stigma of such fundraises as cash needs loom. These types of deals also have consequences for staff, particularly for former employees, who will only have overpriced stock options that become nearly worthless. And such deals can be emotionally brutal. For Turntide, the fundraising has involved “250 investor conversations telling them they have to take big paper markdowns,” said Ryan Morris, the company’s CEO. But that beats running out of cash entirely. “The most important thing is recognizing reality,” Morris said.
In some cases, investors have to put new cash into companies that are at least somewhat distressed to avoid a total wipeout of their stakes, in pay-to-play deals. At farming startup Indigo Ag, last valued at close to $4 billion, investors recently learned that the startup was planning to sell new shares at a 96% discount from its round last year, two people familiar with the matter said. If they refrained from investing, they would lose additional ownership rights that come with preferred stock, one person said. An Indigo spokesperson declined to comment.
Other Silicon Valley private tech stalwarts could be making similar moves. Quora, a startup that has raised nearly $300 million over 14 years, most recently at a $2 billion valuation, two weeks ago filed a corporate charter update that authorizes it to issue only common stock rather than the preferred stock investors previously got. In a cram-down financing, investors who don’t put in new money often lose their designation as preferred shareholders, meaning they no longer get paid out first in a sale or liquidation. Adam D’Angelo, Quora’s CEO, declined to comment.
Plans or discussions for financings at Turntide, Nuro, Indigo and Quora haven’t been previously reported.
Becki DeGraw, a partner at law firm Wilson Sonsini Goodrich & Rosati who advises both startups and venture capital firms, said more of her startup clients have started to ask her for advice on down-round deal structures as cash runs low. She is coaching them on concepts like cram-down financings, as well as penny warrants allowing investors to buy additional shares in a company for a nominal price, which can help offset valuation pain.
“The deals we’re going to see are going to be messier and more complex,” DeGraw said.
Quiet Down Rounds
More than 900 private companies passed the $1 billion threshold between 2021 and 2022, according to Crunchbase, when money was cheaper and public tech stocks soared. Some young startups were able to raise at valuations that were 100 times their low annual revenues.
As stocks fell into a bear market, private valuations mostly remained frozen as founders who had taken advantage of free-flowing investing dollars drew down cash and laid off workers. Some firms cobbled together bridge financings, small checks typically raised between larger rounds, from existing investors or raised debt, maintaining their earlier valuations. There were notable exceptions, such as Stripe’s funding, which lowered its valuation to $50 billion from $95 billion.
“Companies were basically doing everything they could to avoid down rounds,” said Will Ballard, an analyst at Lagniappe Labs, which runs a private markets index for startups.
But startups now are running out of options. Some are quietly raising more-traditional down rounds, in which they sell shares at lower prices than they had previously. Rho Technologies, a New York–based startup that offers corporate cards and expense management software, in May raised $40 million at a $385 million pre-money valuation in a round led by early-stage venture firm M13. That was about 10% lower than Rho’s prior pre-money valuation of $425 million. The down round, which hasn’t been previously reported, came despite the firm’s forecast of a 2.5 times increase in revenue this year, to $50 million, a person familiar with the matter said.
Global Venture Funding In July Was Second-Lowest This Year As Seed Startups Are Hit Hard
August 8, 2023
Global venture funding in July 2023 was the second-lowest monthly total since the reset began more than a year ago, Crunchbase data shows. With the slowdown now in its fourth or fifth quarter, it increasingly looks like the startup ecosystem is undergoing a top-to-bottom reset, from seed through late-stage startups and all the way to the investors that back them.
Global venture funding in July 2023 totaled $18.6 billion — down about 20% month over month, and 38% compared to the $29.8 billion invested in July 2022.
While that’s not the lowest monthly total we’ve seen this year, it’s close: Total funding per month so far in 2023 has ranged from a low of $18 billion in February to a peak of $33 billion in January, per Crunchbase data.
All funding stages — seed, early and late — in July 2023 were down close to a third compared to a year ago.
Notably, seed and early-stage funding hit its lowest amount in a single month since we began tracking the downturn in July 2022. That’s a scary sign for the startup ecosystem as a whole, since earlier funding stages had remained relatively insulated at the start of the downturn.
Now, even those fledgling companies are struggling to raise their early rounds of funding — and we’re also seeing those companies that do raise seed and early-stage rounds struggle to graduate to the later stages.
The seed funding ecosystem is also the most exposed due to the large number of new players and the inherently risky nature of seed-stage investing. After all, it’s widely believed that even in a normal market, 50% to 90% of startups fail.
Startup valuations reset
When the public markets started their downward slide in December 2021, it took the private funding markets a full quarter before registering the need to scale back on valuations.
By the second quarter of 2022, the amount of late-stage funding to startups had come down significantly. Investors signaled they would pivot away from investing in highly valued companies and focus on companies at the earliest funding stages who were years away from exiting.
But, as many startups have seen sales slow in this post-pandemic world, and raising later-stage funding became more difficult, even the previously robust early-stage funding environment started to tighten as well.
By the third quarter of 2022, early-stage funding fell. In the fourth quarter, it became obvious seed funding wasn’t safe either.
The protracted slowdown has continued into 2023. That has given way to concerns that the handoff between investors at each stage is broken as startups find it much more difficult to raise follow-on funding.
Seed- and early-stage startups face a reckoning if they are not able to raise funding on a two-year cycle. As investors cut back, startups funded during the market peak face closure at greater rates.
Arguably, for years the whole seed-stage funding ecosystem grew massively because there was a strong exit market at the end of the pipeline.
Seed investor Sam Lessin predicts that the seed ecosystem will go into “time out” for at least 18 months — if not longer — “until the inventory of dramatically over-marked late-stage private deals got worked through / washed out / expired on the line.”
That could cause the entire seed funding ecosystem to die off temporarily, before being reconstituted, he predicts. That’s because “run-of-the-mill public tech companies just aren’t worth that much it turns out — and if the bulk of so-called unicorns can’t get public / or do and are disappointing, the whole model of seed investing starts to look way way less attractive as an asset class.”
The blush came off unicorn valuations some time ago. And last month, new unicorns hit the lowest count since we started tracking new unicorn counts at the beginning of 2020.
Largest sectors
Perhaps surprisingly, funding to AI companies did not stand out as a sector this past month.
Funding to AI companies was around $2 billion — about 11% of all venture funding — in July. Health care and biotech companies raised around $3.8 billion, renewable energy around $3.1 billion, and financial services companies raised $2.9 billion this past month.
VC Optimism Returning But More Pain Ahead In Their Portfolios
Posted on August 8, 2023 by hunterwalk
Ok, this is the “For VCs, There’s More Pain Coming” post that I promised earlier (while also suggesting it’s actually a GREAT time to start a company). Obvious caveats to my POV here, most specifically: exposure is limited to largely the US/SiliconValley ecosystem, driven by our own portfolio, my friends and co-investors, the funds I’m a LP in, and our institutional LP relationships. But since this is vibes > data anyway, I’ll start with a story from Homebrew’s 2023 Annual Meeting.
Satya and I were having lunch (yummy Chinese food) with our LPAC and the conversation turned to generally “how much more did venture portfolios have to fall before they found their true current value?” That is, for the class of funds institutional LPs tend to back, on average, where was bottom? Each underlying firm has its own ‘valuation policy’ and we can have a separate conversation about the quality of those estimations, but you can generally assume that (a) there’s no real incentive for established VCs to be out of line with their view of reality (this stuff gets approved by accountants) and (b) LPs see this across a variety of managers and are sophisticated enough to apply their own modifiers to the numbers they are provided.
At the time, this is last quarter and the stock market has trended upwards nicely since then (a potential leading indicator of private tech valuations), we all agreed venture portfolios were probably still 25-40% overvalued. That’s a big number, one which if accurate moves many funds to at/below their target return goals for at least the moment! Our estimates were not out of line with new data from top firms like USV who, according to reports, “marked down the value of seven of its funds by nearly 26%.”
What are my major assumptions for why there’s more markdowns to come in the aggregate for the last decade of venture portfolios?
Valuations. The number of startups who raised money beyond the ‘Unicorn’ benchmark grew so dramatically before the 2022 reset that there is just simply farther to fall when many of these fail to grow into their targets, or disappear completely. The capital piled into them also transformed them, asking them to grow faster, spend more, and so on. These mutant unicorns may not recover without dramatic changes to culture and strategy, not just spend.
Fund Sizes Got Too Big. Firms raised too much money. I’m not crying for them – it’s their fault and they’re getting paid hefty management fees even if they’re mediocre investors – but greed and/or competitive pressure (plus an influx of new LPs) caused many VCs to have fund sizes which outpaced their capacity to deploy prudently and their existing strategies.
Restructures, Down Rounds, and Pay to Plays. Whatever gets reported is just the tip of the iceberg. The reality is lots of companies – many of them quite promising – have already undergone, or will be facing, next financings which “clean up” old cap tables. Often not all insiders have the dry powder to protect their positions, or feel the juice isn’t worth the squeeze. Sometimes these are led by outside investors and old ones will just take the impact and walk away. Regardless, even in rounds with no punitive structure, the quickest way to underperformance as a fund is by increasing your ‘dilution before exit’ portfolio model assumptions by 1000-5000 basis points. And that’s what’s happening here.
Soft Acquisition Market. Chilling effect of FTC action on major tech M&A combined with public company shareholders wanting their companies to maintain/grow profitability versus spend on what could be still overpriced assets. As The Information proclaims, “Unicorn Fire Sales Ahead.” At the lower end of the market, acquihires aren’t returning much to the cap tablesand others are carving themselves into pieces to find buyers (and cash).
Many VCs Owned Too Little of Their Portfolio Companies to Begin With. When markets were at their peak the discipline around ownership felt antiquated to some, or at least challenged by the competitive realities. So when great exits again return to the $1b, $3b levels instead of everything being $5b-$50b on paper, it causes a lot of pain. As I wrote about last year, this is a huge (but not unexpected) change to models. Quite simply, on a $10b outcome everyone eats, but on a $1b outcome only concentrated investors see enough back to move the needle and/or those investors who got in early and keep their fund sizes reasonable. The growth in fund sizes plus the decrease is outcome size coupled with ownership challenges is a disaster. When the company exits you’ll get all the ‘congrats’ but you’ll know the DPI doesn’t match up. Let me tell you *every* credible VC fundraise deck I’ve seen this year talks about the importance of ownership concentration.
More Than Average (ALLEGED) Fraud. If not in number of companies, then seemingly in the amount of capital they were able to raise before getting exposed.
Overweighted in Speculative Crypto and Weren’t [Slimy or Smart depending on your POV] Enough To Get Out Before The Shitcoin Collapse.
So yeah, it’s gonna be a tough vintage of returns for many but hopefully healthy for our industry. Lower performing VCs will disappear faster and new entrants will differentiate themselves. Funds will get rightsized, which helps better align investors and founders in what defines a successful outcome. And fascinating new advances (and needs) in AI, climate, biology, etc are driving tech-IP driven startups.
AI of the Week
OpenAI Shoots To No. 1 Spot On Private Cloud Startup Ranking, Underscoring AI’s Sudden Ascendency
August 9, 2023
OpenAI debuted at the No. 1 spot on the Cloud 100 Benchmarks Report 2023, a ranking of the top 100 private cloud companies conducted every year by Bessemer Venture Partners. OpenAI’s rise to the top of the ranking underscores artificial intelligence’s shockingly fast ascendency in the tech world. Besides the ChatGPT owner, four other native AI companies joined the index this year: Anthropic (No. 73), Midjourney ( No. 85), Hugging Face (No. 98) and DeepL (No. 100).
Bessemer, one of the most active venture investors in the cloud sector, taps a panel of judges each year to evaluate companies on four criteria for inclusion on the Cloud 100: market leadership, financial metrics, valuation and team.
“It was the hardest year ever to make the cloud 100 list,” Mary D’Onofrio, a partner and co-founder of the growth practice at Bessemer, told Crunchbase News. “The minimum valuation required to make the list was close to $1.5 billion.”
And in contrast to prior years, only one company — Figma — exited from last year’s leading 100 cloud company list, even though Adobe‘s $20 billion acquisition of the design collaboration startup has not yet closed. (Figma is one of the highest acquisition multiples of all time at 50x revenue.)
AI is clearly a leading theme for this year’s index. Of the 100 companies on this year’s index, 55 have launched a generative AI product or feature in the last eight months, and 70 use AI or machine learning in their product.
However, design, collaboration and productivity — a sector impacted by AI — led with the highest collective value, at $110 billion. That displaced last year’s most-valuable sector, fintech, after some companies saw their valuations decline. That includes Stripe, which raised a down round, and Checkout.com, which reportedly cut its internal valuation by 70%.
Cloud security company Wiz moved up the most on the index — 67 spots to No. 15. Wiz took 18 months to scale from $1 million to the $100 million in revenue announced earlier this year.
IPO predictions
The report predicts the IPO market will also open up soon — D’Onofrio predicts it will do so in the first half of next year — with the top 10 companies on the index most likely to go public first.
The cloud 100 companies were valued collectively at $654 billion, down 11% from last year’s list. That marks the first annual decline on the index since 2016.
The average value of the companies on this year’s list also fell, from $7.4 billion in 2022 to $6.6 billion this year.
Average growth has slowed from 100% to 55% on average year over year. The top quartile also slowed from 120% to 70%. Revenue multiples were calculated from companies that raised in the past year, which include 17 companies valued at 26x revenue. That’s still high compared to public cloud companies, at around 8x revenue. The public cloud index is growing at a slower rate, on average 22%.
“On a growth adjusted basis, it’s actually a lot more rational and more in line with the public markets than one would think,” said D’Onofrio. “You’re paying for the outsize revenue growth in this environment, combined with your view of what a company can become.”
Generative AI Is Making Companies Even More Thirsty for Your Data
The outcry over Zoom's tweak to its data policy shows how the race to build more powerful AI models creates new pressure to source training data—including by juicing it from users.
PHOTOGRAPH: JONATHAN KNOWLES/GETTY IMAGES
ZOOM, THE COMPANY that normalized attending business meetings in your pajama pants, was forced to unmute itself this week to reassure users that it would not use personal data to train artificial intelligencewithout their consent.
A keen-eyed Hacker News user last week noticed that an update to Zoom’s terms and conditions in March appeared to essentially give the company free rein to slurp up voice, video, and other data, and shovel it into machine learning systems.
The new terms stated that customers “consent to Zoom’s access, use, collection, creation, modification, distribution, processing, sharing, maintenance, and storage of Service Generated Data” for purposes including “machine learning or artificial intelligence (including for training and tuning of algorithms and models).”
The discovery prompted critical news articles and angry posts across social media. Soon, Zoom backtracked. On Monday, Zoom’s chief product officer, Smita Hasham, wrote a blog post stating, “We will not use audio, video, or chat customer content to train our artificial intelligence models without your consent.” The company also updated its terms to say the same.
Those updates seem reassuring enough, but of course many Zoom users or admins for business accounts might click “OK” to the terms without fully realizing what they’re handing over. And employees required to use Zoom may be unaware of the choice their employer has made. One lawyer notes that the terms still permit Zoom to collect a lot of data without consent. (Zoom did not respond to a request for comment.)
The kerfuffle shows the lack of meaningful data protections at a time when the generative AI boom has made the tech industry even more hungry for data than it already was. Companies have come to view generative AI as a kind of monster that must be fed at all costs—even if it isn’t always clear what exactly that data is needed for or what those future AI systems might end up doing.
The ascent of AI image generators like DALL-E 2 and Midjourny, followed by ChatGPT and other clever-yet-flawed chatbots, was made possible thanks to huge amounts of training data—much of it copyrighted—that was scraped from the web. And all manner of companies are currently looking to use the data they own, or that is generated by their customers and users, to build generative AI tools.
Zoom is already on the generative bandwagon. In June, the company introduced two text-generation features for summarizing meetings and composing emails about them. Zoom could conceivably use data from its users’ video meetings to develop more sophisticated algorithms. These might summarize or analyze individuals’ behavior in meetings, or perhaps even render a virtual likeness for someone whose connection temporarily dropped or hasn’t had time to shower.
The problem with Zoom’s effort to grab more data is that it reflects the broad state of affairs when it comes to our personal data. Many tech companies already profit from our information, and many of them like Zoom are now on the hunt for ways to source more data for generative AI projects. And yet it is up to us, the users, to try to police what they are doing.
“Companies have an extreme desire to collect as much data as they can,” says Janet Haven, executive director of the think tank Data and Society. “This is the business model—to collect data and build products around that data, or to sell that data to data brokers.”
The US lacks a federal privacy law, leaving consumers more exposed to the pangs of ChatGPT-inspired data hunger than people in the EU. Proposed legislation, such as the American Data Privacy and Protection Act, offers some hope of providing tighter federal rules on data collection and use, and the Biden administration’s AI Bill of Rights also calls for data protection by default. But for now, public pushback like that in response to Zoom’s moves is the most effective way to curb companies’ data appetites. Unfortunately, this isn’t a reliable mechanism for catching every questionable decision by companies trying to compete in AI.
In an age when the most exciting and widely praised new technologies are built atop mountains of data collected from consumers, often in ethically questionable ways, it seems that new protections can’t come soon enough. “Every single person is supposed to take steps to protect themselves,” Havens says. “That is antithetical to the idea that this is a societal problem.”
Money on Autopilot: The Future of AI x Personal Finance
by Olivia Moore and Anish Acharya
Seventy-seven percent of Americans report feeling financially stressed, making consumer personal finance an attractive market for new builders. And yet, the space is home to dozens (if not hundreds) of startups that have helped their users, but haven’t been able to fully transform their financial lives.
Why? There’s historically been a massive gap between consumer expectations for personal finance products, and what those products can accomplish. Most digital personal financial managers (PFMs) have focused on surfacing insights about your money. At best, they help you learn about new products or behaviors—which could help, if you go to the time, effort, and expense of following through (and maintaining / rebalancing over time).
Personal finance is a complicated and emotionally charged topic, often in negative ways. Several studies have found that most people would rather talk about any other topic—including sex or death—than their finances! Consumers don’t want to spend more time thinking about their financial situation. They want someone to fix it for them and, even better, to keep them on track over time.
Money on Autopilot
Thanks to generative AI, the much-discussed topic of “self-driving money” finally has a chance to achieve its potential. Imagine a platform that can move your money to optimize your balance sheet. In the past, this wasn’t possible from a technical standpoint, as products were stuck in “read only” mode. They could generate information or analysis, but couldn’t take action on your behalf—which is arguably the most important step.
Post-generative AI, we’re in a new world for consumer financial platforms. LLMs, and specifically multi-modal prompts like GPT-4, can process and output both text and images. This enables consumer robot process automation (RPA), which will allow fintech apps to operate on a user’s behalf. This massively opens up the universe of potential user interactions with personal finance products in terms of both inputs (what the product can analyze) and outputs (what the product does for the user).
As one example: Google’s Bard is able to ingest a screenshot of your investment account balance, “read” the starting and ending values (as well as deposits and withdrawals), and calculate your investment returns benchmarked to the broader market. Bard has only been live for four months, so we expect this functionality to get more and more sophisticated over time.
As a result of this, we expect to see startups finally deliver on the vision of financial automation, with products that serve as “autopilots” to help consumers:
Save and spend
Make investments
Plan for retirement
Manage debt
Prepare/file taxes
And more!
Alongside autopilots that optimize your assets within a category (e.g., analyze and rebalance your stocks across brokerage accounts), we may even see the rise of the first great financial super app in the U.S. This would serve as an autopilot across all of these product categories, allowing for a 100% “hands free” management experience, which can route money between your existing apps and accounts. Essentially, this becomes an AI accountant and wealth manager for the masses, which not only sets you up for success, but automatically re-allocates your money as your life changes.
Importantly, consumers wouldn’t have to change providers—they could continue to use separate apps for investing, saving, spending, etc., with the autopilot providing an optimization layer across these apps.
Successful companies here will look different than anything we’ve seen in consumer finance before. Most notably, they won’t rely on consumer engagement to deliver value. In fact, the best products will have incredibly fast and smooth onboarding and a “set it and forget it” motion, with success measured on how much of their wallet or portfolio a user hands over to them over time…..
Apple seeks to bolster expertise in generative AI on mobile devices
Tech giant recruits staff to compress existing language models to run efficiently on iPhones and iPads rather than the cloud
Madhumita Murgia in London and Patrick McGee in San Francisco, AUGUST 5 2023
Apple is bulking up its expertise in generative AI to adapt it for iPhones and iPads, as the world’s biggest company by market value seeks to take advantage of the technology that has taken the industry by storm this year.
The Cupertino-based tech giant is hiring for dozens of roles across offices in California, Seattle, Paris and Beijing that will work on large language models or LLMs — software that can produce plausible text, images or code in response to simple prompts.
All the jobs were advertised between April and July and indicated that Apple was working on “ambitious long-term research projects that will impact the future of Apple, and our products”.
Multiple teams at Apple, such as the Machine Intelligence, Neural Design (MIND) group, are recruiting researchers and engineers for jobs ranging from fundamental research on LLMs in its Paris lab, to compressing existing language models so they can run efficiently on mobile devices, rather than in the cloud.
While rivals such as Microsoft-backed OpenAI and Google have been quicker to release generative AI products such as chatbots and productivity assistants, Apple’s job ads indicate the company is expanding efforts to bring cutting-edge technologies like LLMs specifically to mobile.
This remains a huge technical challenge not yet solved by Apple’s competitors, but will be vital to the iPhone-maker’s core businesses in device sales and associated services.
On a call with investors on Thursday, Apple chief executive Tim Cook called AI and machine learning “core, fundamental technologies that are integral to virtually every product that we build”. Apple’s research and development spending for the third quarter was $3.1bn higher than at this point last year, which Cook partly attributed to generative AI, saying Apple was “investing a lot” over coming months.
The company has thus far talked cautiously about enhancing existing features like autocorrect and animations of photos, using machine learning.
“Their short-term goal is to integrate [generative AI] with existing products to make it productive and useful immediately,” said Lukasz Olejnik, an independent researcher and cyber security consultant who works with big tech companies.
The benefits of running AI software on phones — without the need for an internet connection or to send data to the cloud — are that apps can run more quickly and allow a user’s data to be processed in a more secure and private way. Operating AI on mobile devices would “have superior privacy protection properties,” Olejnik said.
One ad, for instance, says it seeks a senior software engineer to “implement features that compress and accelerate LLMs in our on-device inference engine”, referring to AI operations on mobile rather than on the web.
Another, from July 28, states that the company wants to bring “state of the art foundation models to the phone in your pocket, enabling the next generation of ML-based experiences in a privacy-preserving way”…..
AI Won’t Replace Humans — But Humans With AI Will Replace Humans Without AI
August 04, 2023
Summary. Karim Lakhani is a professor at Harvard Business School who specializes in workplace technology and particularly AI. He’s done pioneering work in identifying how digital transformation has remade the world of business, and he’s the co-author of the 2020 book Competing in the Age of AI. Customers will expect AI-enhanced experiences with companies, he says, so business leaders must experiment, create sandboxes, run internal bootcamps, and develop AI use cases not just for technology workers, but for all employees. Change and change management are skills that are no longer optional for modern organizations.
Just as the internet has drastically lowered the cost of information transmission, AI will lower the cost of cognition. That’s according to Harvard Business School professor Karim Lakhani, who has been studying AI and machine learning in the workplace for years. As the public comes to expect companies that deliver seamless, AI-enhanced experiences and transactions, leaders need to embrace the technology, learn to harness its potential, and develop use cases for their businesses. “The places where you can apply it?” he says. “Well, where do you apply thinking?”
For this episode of our video series “The New World of Work”, HBR editor in chief Adi Ignatius sat down with Lakhani, author of Competing in the Age of AI: Strategy and Leadership When Algorithms and Networks Run the World, to discuss:
How executives and regular employees can (and must) develop a digital mindset
Change management as a critical skill that must be in the DNA of any successful organization
The shapes AI may take in the near and far future
“The New World of Work” explores how top-tier executives see the future and how their companies are trying to set themselves up for success. Each week, Ignatius talks to a top leader on LinkedIn Live — previous interviews included Microsoft CEO Satya Nadella and former PepsiCo CEO Indra Nooyi. He also shares an inside look at these conversations —and solicits questions for future discussions — in a newsletter just for HBR subscribers. If you’re a subscriber, you can sign up here.
ADI IGNATIUS:
Karim, welcome to the show.
Karim Lakhani:
So glad to be here with you today, Adi.
ADI IGNATIUS:
You co-wrote a piece for us a few years ago, and it’s reflected in your book, where you say machine learning has basically changed the very rules of business. That’s a big statement. What do you mean by that?
….
Cloudflare as a Leader in AI? + Q2 Results
August 4, 2023 · by D/D Advisors · in Networking. ·
For the past six months, the investing world has been on the hunt for investible stocks with AI exposure. Surely, they ask, there must be something other than Nvidia benefitting from all this AI spending? The large cap semis universe offers varying degrees of “AI-ness” from AMD with its new M100 GPU, to Intel who maybe has something, to Qualcomm with lots of AI capabilities but no clear way to monetize them. There are also all kinds of wafer fabrication equipment (WFE) and small cap names with AI exposure, all of which are covered well by Fabricated Knowledge and Semi Analysis (seriously, go subscribe to those). But in the end, this is not a big list.
Recently, Josh Wolfe of Luxe Capital appeared on Bloomberg’s Odd Lots podcast. Buried deep in an interesting conversation he mentions that Cloudflare is an unsung winner in AI. This piqued our interest as Cloudflare is one of those companies that we have been keeping an eye on for many years. They seem to sit at an important spot with their globe-spanning network, with equipment in over 200 cities, coupled with a compelling growth engine that seems to be executing incredibly well.
Cloudflare provides a huge array of services for connecting websites, data stores, workers and customers all over the world. They essentially provide a networking service sitting between data centers/the Cloud and all the users at the edge. We will save a deeper dive into their business for some future post, but the more we look at the company the more interesting it seems. Their services are immensely useful in connecting all the pipes that make up the Internet. In fact, one of the most surprising things about Cloudflare is that no one else built their business before them. For those who remember the dot.com Bubble of the 90’s, a number of companies, like Akamai and Limelight, arose to provide functionality to distribute content closer to the eyeballs on the Internet. Translate those terms from 90’s speak to modern day terminology and those companies all had early versions of what Cloudflare offers today. But somewhere along the line, those companies faded away or went far off course, leaving room for Cloudflare to fill the gap just at the time when everyone moved their workloads to the Cloud. If nothing else, Cloudflare is worth following for all the insight they provide into how the Internet actually works at a fairly tactical low level.
The company reported earnings yesterday, and their results were fairly healthy. Q2 revenue of $305 million hit consensus and EPS of $0.10 was three cents ahead. They guided revenue to $330.5 million ahead of consensus of $329.4 million and EPS one cent ahead of $0.09 expectation. They also bumped up their full year estimates to reflect the good quarter. One interesting facet of the company’s reports is that they guide revenue very precisely, Q3 was guided to $330 million to $331 million, the full year was guided to $1.283 billion to $1.287 billion. Anyone who guides with that tight a band has a deeply recurring revenue base, but even other SaaS companies have a guidance range that is an order of magnitude larger than Cloudflare’s.
One of the confounding things about Cloudflare is that they do not neatly fit into industry categories. They provide software, but they are not really a software company. Most of their competitors are actually hardware companies selling point solutions for specific functions like security or application acceleration, and they are definitely not a hardware company. Probably their truest comparables are AWS, Azure and Google Cloud, but no one likes to say that out loud as it makes their position seem untenable. Despite this, the company plugs some important holes and solves some significant problems for anyone building globally accessed software.
That all being said, what interested us most on their call was management’s commentary about AI. First, we appreciated their modesty about the term AI which has seen multiple generations of definitions. More critically, they spoke about AI as an important opportunity for the company, echoing Wolfe’s comments above. As we have noted, the big opportunity for anyone touching AI will be the inference market, and Cloudflare has a compelling story here. Like us, they note that the only way for the economics of AI inference to work will be to distribute a significant portion of the workload on to edge devices (i.e. users’ computers and smartphones). This means there will be a significant need to move a lot of data around the network. Enterprises building their own AI models will need to distribute those models close to the workers actually running inference queries against it. Consumer devices will need to receive regular updates as models improve. All of this will probably work best when run over Cloudflare’s network, which can keep all the data synced up and secure. Moreover, the big motivation for enterprises running their own models will be to allow them to keep control of all that data, the notion of data sovereignty. Cloudflare is probably one of the leaders in providing such sovereignty and is a critical aspect in their competitive differentiation from the hyperscale public cloud providers.
News Of the Week
Yaccarino Says Advertisers are Coming Back to X in First Major Interview as CEO
Published Aug. 10, 2023
By Andrew Hutchinson - Content and Social Media Manager
New X chief Linda Yaccarino says the company is close to breaking even, while the re-brand to X has been popular among users, according to internal insights.
Yaccarino noted this in a new interview with CNBC, in which the recently anointed X chief was questioned about the automony of her role under X owner Elon Musk, her plans for revitalizing the platform’s ad business, and her further vision for the app.
And while a lot of Yaccarino’s statements were exactly what you would expect from somebody who’s trying to pitch the platform as a key consideration for advertisers, Yaccarino did make some interesting notes about their progress, and what may be coming next.
The main focus of the interview was on the re-brand and how that’s impacted the business.
Yaccarino seems confident in the new direction, explaining that X encompasses more opportunities than the previous moniker.
As per Yaccarino:
“The rebrand represented really a liberation from Twitter, a liberation that allowed us to evolve past a legacy mindset and thinking, and to reimagine how everyone, how everyone on Spaces who’s listening, everybody who’s watching around the world, [how] it’s going to change how we congregate, how we entertain, how we transact all in one platform.”
That’s pretty much in line with the bombastic description Yaccarino recently shared to X about the new vision for the app, which had many rolling their eyes at the corporate-speak.
I mean, Yaccarino is the chief of the company, and an expert in media messaging, so this, really, is what you would expect. But there’s a distinct vagueness to these terms, which all sound good when spoken, but are fairly hollow in substance.
Still, Yaccarino was keen to highlight the evolution of the app, and the new opportunities:
“Experiences and evolution into long-form video and articles, subscribe to your favorite creators, who are now earning a real living on the platform. You look at video, and soon you’ll be able to make video chat calls without having to give your phone number to anyone on the platform.”
All of these, Yaccarino says, form the basis of what X is all about, in differentiation to Twitter, though most are pretty much in line with the previous Twitter experience, and were even enacted, in some form, under previous Twitter management.
So as yet, it’s not some huge change in direction. But it is early, especially for Yaccarino herself, who only took the job three months ago, after more than a decade working for NBCUniversal.
Five months ago Crunchbase vowed to try and avoid layoffs. Since then A.I. has changed everything
August 4, 2023 at 4:46 AM PDT
Five months ago, Crunchbase CEO Jager McConnell was telling me over Zoom how he had been cutting costs across the company: They had canceled the holiday party. Hiring was frozen. But no layoffs—that would be a last resort.
Obviously, a lot has changed since the end of February—and especially at Crunchbase, the Mayfield and OMERS Ventures-backed startup that has become a go-to provider for data and fundraising information on private companies and investors. Three weeks ago, McConnell posted on LinkedIn that he had laid off at least 49 of the 241 staffers Crunchbase had in March—predominately in its go-to-market teams, including sales and marketing. (In total, 63 employees were impacted—as McConnell says some voluntarily left when he warned employees layoffs were coming a month prior, and others didn’t want to be added to the list he shared on his social media profile.)
So what happened? In the five months since our last conversation, McConnell says he had come to a major realization: Because of recent developments in artificial intelligence (so-called generative A.I.) Crunchbase’s executive team had determined at an offsite that the company needed to hire more data scientists and engineers and rethink every layer of its business—how they sourced data, how they interpreted the data, and then how they delivered it to the end user—”or we would have been left behind,” he says.
“It’s an existential threat to your business if you don’t think of things as in an A.I.-first world,” McConnell says, equating the changes they are making to building a “new department in a lot of ways from what we had before.”
Right now, Crunchbase pulls data from its platform users, from 4,000 data partnerships, and from A.I. tools Crunchbase’s data teams have already built, which perform tasks like identifying new companies that should be added to Crunchbase or verifying whether a company is still in business. Now Crunchbase is planning to double its data engineering teams—from two to four—with plans to hire machine learning engineers, data engineers, data scientists, data project managers, and data engineering managers. To be able to afford those new staffers, McConnell says they had to let other staffers go.
Crunchbase is currently developing a model to write short company descriptions for its site—which, at the moment, costs about 50 cents a pop for humans to put together, making it the most expensive task for their software. McConnell says they are also planning to build a new search function for more intuitive user interactions with Crunchbase data, and even some new predictive capabilities, such as when companies will be fundraising soon, or which companies people are talking about at the moment (or should be).
Crunchbase hired Chief Product Officer Megh Gautam at the end of May—the focus being how the company can better push into the direction of A.I.
Crunchbase is likely not an outlier with the generative A.I. wave crashing across the entire business ecosystem right now: Every company is strategizing how they can utilize large language models to transform the way their business functions. But it’s a stark reminder of exactly what people have been warning about—that the newfound obsession with A.I. will come at a cost for preexisting jobs. In coming months, I’m expecting to see many more iterations of what Crunchbase is doing. Likely more layoffs, too—not because of pressure from investors to free up cash, but because businesses are rethinking the very structure of how they run.
That’s yet another burden to the tens of thousands of people who have already been laid off due to rampant cost cutting across the tech and startup ecosystem.
“I wish I had a better sense of this happening and how it was going to impact us,” McConnell says. “Maybe that would have changed our hiring decisions earlier…[This was certainly hard] on our employees and the people that were impacted, and that’s on me, and that means I was making mistakes.”
WeWork’s going concern warning is a reminder that VC and low-margin business don’t mix
Alex Wilhelm, @alex / 8:30 AM PDT•August 9, 2023
Image Credits: MANDEL NGAN/AFP/Getty Images / Getty Images
There’s no point in being rude to WeWork at this juncture. The company’s market cap has fallen to around $130 million, it has billions of dollars in debt, and it said recently that it may struggle to stay in business as its cash balance dwindles. WeWork sees several avenues to right the ship before it runs out of cash: Reduce rent and tenancy costs, limit user churn, reduce its overall cost basis and raise new capital.
It is not clear how much relief those endeavors will bring, but WeWork is certainly not going out without a fight. Its debt restructuring effort earlier this year is more evidence of that intent.
As we enter what could be the last few months of WeWork, we can draw several lessons from its Icarus-esque rise and fall. You could argue that WeWork is a warning against granting founders too much control for too long: Its founder was famed for his ability to sell and raise capital, but lax controls failed to prevent delusion from replacing ambition at the company. You could also make the case that WeWork became too complicated a financial entity for its own good.
But here’s the lesson I want to take away from WeWork’s saga: Venture capital can be excellent for quickly scaling technology startups, but the model is not a good fit for lower-margin businesses.
Costs, losses, weights
In its most recent quarter, WeWork reported revenue of $844 million, up 3.6% from a year ago. The company improved its bottom line, too, narrowing its net loss to $397 million from $635 million, and shrinking adjusted EBITDA losses to $36 million from $134 million.
Those figures, however, did nothing to offset the fact that the company still had a free cash flow deficit of $646 million in H1 2023. That kind of cash burn is a tough obstacle to overcome for a company that’s worth less than a quarter of its free cash flow deficit from just the first two quarters of the year.
WeWork’s description of its liquidity isn’t encouraging, either:
As of June 30, 2023, the Company had $205 million in cash and cash equivalents, including $46 million held at its consolidated VIEs, and $475 million in delayed draw note commitments, resulting in total liquidity of $680 million. The Company issued $175 million of the delayed draw notes in July 2023.
It’s no surprise that the company’s “losses and [its] projected cash needs, which have been impacted by the recent increases in member churn, combined with [its] current liquidity level [leave] substantial doubt . . . about the Company’s ability to continue as a going concern.”
Did it have to be this way? I don’t think so. WeWork clearly had an idea — there’s a good book about that — that resonated with its customer base. However, in its pursuit of ever-larger checks, the company prioritized revenue growth over the health of its business. The blame for WeWork getting out of shape so fast does not rest solely with investors per se. I’d hazard the fault lies with everyone who was involved.
When you pour capital into a business with limited gross margins to help it increase revenue while accreting long-term liabilities, you can spend your way to a point of no return. Why do margins matter, then? Because all the money that WeWork spent on growth has left it with a revenue base that is simply not profitable; what all that fundraised money bought is actually not worth that much.
In Q2 2023, here’s how the company’s gross profit stacked up:
Revenue: $844 million.
“Location operating expenses — cost of revenue” sans depreciation and amortization: $725 million.
Gross profit sans depreciation and amortization: $119 million.
For a company that has selling, general and administrative costs of $150 million before taking into account debt costs, it’s clear that WeWork remains far from even reaching operating profitability. And we may be being too generous here by not counting the depreciation and amortization costs in our math. Indeed, in its Q1 2023 report, WeWork said that its very non-GAAP “building margin” metric came to –$20 million, inclusive of depreciation and amortization, and +$120 million without.
It’s very cool that WeWork has a revenue run rate of $3.38 billion, but if that revenue costs a lot to generate, it’s hard to arrive at the company’s value — apart from knowing that it is not much. Investors seem to agree with that estimation, with the company’s shares down 22% to $0.16 this morning following the going concern warning and the Q2 results.
Startup of the Week
A better internet for readers
What you read matters; so does where you read it
CHRIS BEST, HAMISH MCKENZIE, AND JAIRAJ SETHI
AUG 10, 2023
The internet revolutionized reading, but instead of a utopia, it has delivered a mess. The main places where we read online today are cacophonic, stressful, and milking our minds for ad dollars. There are great reading products, but they are niche, and the tech giants seem to have lost interest in making them since the demise of Google Reader. Instead, we are left to contend with a fusillade of pop-ups and a Big Social-dominated media economy that is making us angry and stupid.
But the internet still holds tremendous promise for readers. It remains true that it democratizes access to news, information, and culture, so that anyone, anywhere can enjoy work that in previous generations was reserved for the privileged. Its powerful networking tools can still produce amazing communal reading experiences, and its viral dynamics can take memes to the moon. The trick is to weave these elements together with an economic system that incentivizes quality over superficial engagement, and trusted connections over tribal tantrums. No one has quite mastered that art.
Perhaps it is naive, but we at Substack haven’t given up hope. We think it is still possible to harness the internet’s powers to create a better world for readers. We can see a future where reading online is a pleasure, with fast-to-load posts, clean and uncluttered pages, and simple navigation. We believe in a business model that gives readers the power to help shape culture by directly supporting the writers and work they most value, leading to an incentive system that rewards quality and applies upward pressure for excellence in even the smallest of niches. We think that reading can be social without being distracting. And we bet that trusted peer recommendations can drive a discovery system that helps the world’s best readers find the world’s best work—no matter where it comes from.
So here’s what we’re going to do: We are going to make a beautiful and exceptional place for readers that extends the platform we’ve built for writers and uses the best of technology to get the best for culture.
Over the coming months and years, we’ll be adding features and evolving our reading apps so that they feel increasingly useful and fun. You’ll not only have a quiet place to read but also somewhere to hang out with the smartest people you know. It’ll be a space where you can establish a home for your cultural interests and build an audience even if you don’t have a publication. And it will all be tied together in a network of meaningful connections—represented by subscriptions—that prioritize trust over time spent or eyeballs captured.
It will be the best reading experience on the internet.
It has been clear for a while now who a Substack writer is. Writers have a direct and obvious relationship with the platform, the place where they publish, grow, and get paid. But until recently, few people would have described themselves as a Substack reader. Over time, we think this dynamic will change. Even while the writer-reader relationship will remain sacrosanct here, it will become ever clearer that there is such a thing as a Substack reader.
A Substack reader is someone who might be on the verge of opting out of online media because of their aversion to the toxicity of their social feeds. It’s someone who wants high-quality news and culture. It’s someone who’s willing to consider a range of sources, even ones that challenge their assumptions. It’s someone who wants to find a way to be online with dignity.
In creating a home for these readers, we can also better serve writers. Substack will be a gathering place for people who have a propensity to pay the writers they trust—and that’s a group that writers new and established can address when it comes to building their own audiences and influence.
Readers might be the internet’s most important constituency, and yet they have been treated like paupers. The companies that dominate today’s cultural sphere seem to consider the reader’s mind only insofar as it can be tricked into ceding its bounty of attention—the black gold of the “Information Revolution.” What a disappointing revolution it has been. We thought we were going to live in the Library of Alexandria; instead we’re wading through its sewers hoping to find wisdom on wet pages pasted to the walls.
The cultural challenges that we face as readers affect how we think and act together, and our response to them will determine how we confront the many problems we face as a society. When the author
launched Story Club on Substack, he wrote: “[H]ow we tell and receive stories is central to how we think, which, in turn, determines how well (how lovingly, how fully) we live.”
What we read matters.
X of the Week
Twitter for Sale by X
Elon Musk to auction off Twitter memorabilia from San Francisco HQ
Platform rebranded X last month lists 584 lots including Twitter signs, a wooden bird table and outsized bird cages
Hibaq Farah UK technology reporter
Thu 10 Aug 2023 11.05 EDT
Elon Musk is to auction Twitter memorabilia from its San Francisco headquarters following the social media platform’s switch to the name X, including its former bird logo from the side of the building.
The billionaire rebranded the site late last month with a new logo, an X, replacing its distinctive bird symbol. On announcing the move, Musk tweeted: “And soon we shall bid adieu to the Twitter brand and, gradually, all the birds.”
