Layoffs aren’t over, Chinese AI, and Ad struggles are Reel

22 October 2023 | Issue #12 - Mentions $MSFT, $GOOG, $NVDA, $BIDU, $META, Tiktok, $AMZN

Welcome to the twelfth edition of Tech takes from the cheap seats. This will be my public journal, where I aim to write weekly on tech and consumer news and trends that I thought were interesting.

Let’s dig in.

It’s still the year of efficiency

It looked like the tech job layoff cycle was nearing its end, as there have been a consistent decline in employee layoffs month by month since July. But hold off celebrating just yet, because after this weeks announcements from a couple big tech companies, October is now exceeding the month before with still over a week to go before month end. Linkedin and Alphabet announced further cuts, of which the latter is targeted at non-core units.

Microsoft's LinkedIn said on Monday it would lay off 668 employees across its engineering, talent and finance teams in the second round of job cuts this year for the social media network for professionals amid slowing revenue growth.

The cuts, which affect more than 3% of the 20,000-strong staff, add to the tens of thousands of job losses this year in the technology sector in the face of an uncertain economic outlook. "While we are adapting our organizational structures and streamlining our decision making, we are continuing to invest in strategic priorities for our future and to ensure we continue to deliver value for our members and customers," LinkedIn said in a blog on Monday. LinkedIn makes money through ad sales and by charging for subscriptions to recruiting and sales professionals who use the network to find suitable job candidates.

In the fourth quarter of its fiscal 2023 year, LinkedIn's revenue increased 5% year-on-year, compared to 10% in the previous quarter.

Microsoft has cited a slowdown in hiring along with a decline in advertising spending as headwinds for LinkedIn, although it continues to add new members to its community of 950 million

Google and its parent company Alphabet has quietly laid off staff, spokespeople for the conglomerate said, as tech firms continue to tighten their belts following a marked slowdown in business growth.

Divisions such as Google News, Alphabet’s healthcare analytics subsidiary Verily and self-driving car unit Waymo recently laid off staff, including this week, the spokespeople said. The moves indicate Alphabet is determined to reduce costs even after cutting 6% of staff, 12,000 workers, in January in its first mass layoff in more than a decade.

Google has emphasized that the latest round of cuts are much smaller and isolated to particular teams, although the company would not confirm the specific number of employees affected. But at Google News, for example, roughly 45 employees were laid off, including staff focused on regulatory compliance, a spokesperson for the Alphabet Workers Union said.

As it relates to LinkedIn, the cuts are slightly less than 716 that it did in May this year, though it could imply that the business has yet to see stabilisation. There is a little irony in these cuts. Almost half of the company’s revenue comes from Talent Solutions, which helps companies source and hire employees. Seeing further cuts here may suggest that the employment market has yet to bottom (of course the official unemployment numbers of 3.8% are still well below long term averages of 5.7%, but I assume that has more to do with labor force composition and LinkedIn leaning heavier in white collar services).

As a shareholder in both companies, it’s nice to see efficiency is still top of mind. With all the euphoria surrounding AI in these last few months I was starting to get a little concerned that excessive spending may be on the horizon.

The final nail in China’s AI race?

Back in August, I wrote about how the US was holding back the Chinese tech giants in AI by only allowing them to use the slower Nvidia GPUs (A800 & H800, not the A100 & H100). This week brought additional news indicating even more stringent export restrictions

The Biden administration is tightening restrictions on China’s ability to buy advanced semiconductors, fueling friction with U.S. businesses that sell to the vast Chinese market.

The Commerce Department said Tuesday that it would significantly constrict exports of artificial-intelligence chips, making it tougher for U.S. companies Nvidia and Intel to sell existing products in China —or to introduce new chips to circumvent the rules.

The move aims to close perceived loopholes in export controls announced a year ago, which themselves had faced strong opposition from the global semiconductor industry and escalated tensions with Beijing.

The goal, Commerce Secretary Gina Raimondo said, is to limit China’s “access to advanced semiconductors that could fuel breakthroughs in artificial intelligence and sophisticated computers.” The chips are critical to Chinese military applications, she said, a nod to concerns that the U.S. could fall behind China in key defense technologies.

I recommend reading this from Ben Thompson at Stratechery for more context/analysis. This article from the FT also does a good job digging into the dilemma the Chinese tech giants are facing.

Alibaba and Baidu’s processors have become frontrunners in China’s efforts to create domestic alternatives to US maker Nvidia’s sophisticated products, with the chips at present manufactured at TSMC and Samsung plants, said four people close to the groups’ design projects.

The processing speeds of their most advanced AI chips fall within new thresholds unveiled by Washington this week as part of an update to its chip export controls, the people said, putting their partners in contravention of the rules if they manufacture them for Chinese clients.

The tightening will also force Silicon Valley-based Nvidia to halt shipments to China of two processors that the company had tailor-made to comply with earlier export controls, according to a statement from Nvidia this week.

Collectively, the restrictions mean Chinese tech groups will have to turn to AI chipsets similar to Nvidia’s V100, which was released in 2017 and has since been discontinued, in order to train and run generative AI models, analysts said. Since the V100 was released, chips have become significantly more advanced, enabling the creation of OpenAI’s ChatGPT.

The US move poses “an existential challenge” to China’s efforts to catch up with AI development at OpenAI and other American companies, said one chip consultant in Beijing.

Washington’s controls, which extend to foundries in Taiwan and South Korea contracted to make chips for Chinese groups, are made possible by the vast amount of American hardware and software embedded in the semiconductor supply chain. China’s domestic alternatives, including partially state-owned SMIC, are several generations behind in chip manufacturing technology.

While China’s largest tech groups have stockpiles of AI chips, the controls will eventually make training AI models in the country more expensive and time-consuming than for their US counterparts, analysts said. Bernstein senior analyst Boris Van estimated that relying on chips similar to Nvidia’s V100 would at least double data processing costs.

“Once the existing stash of chips is exhausted, Chinese AI firms would struggle to improve their models,” said Phelix Lee, an analyst at Morningstar.

Big Chinese tech groups, including Alibaba, Baidu, ByteDance and Tencent, have purchased more than $5bn worth of Nvidia chips in recent months, the Financial Times reported in August, but most of these orders have not been delivered, according to several people familiar with the situation.

“The supply is terrible,” said a Beijing-based AI entrepreneur desperate for Nvidia’s processors, noting that the company was months behind in deliveries. Washington has given Nvidia and other chip companies a grace period of about one month to fulfil orders to China.

“Whatever portion cannot be fulfilled in the grace period will have to be cancelled,” said Charlie Chai, a Shanghai-based analyst at 86Research.

On the same day these new restrictions were announced, Baidu unveiled its competitor to GPT-4 - ERNIE 4.0 (about 7 months later).

The new ERNIE Bot “is not inferior in any aspect to GPT-4,” Baidu’s billionaire CEO, Robin Li, told an audience at its annual flagship event.

Speaking onstage, Li showed how the bot could generate a commercial for a car within minutes, solve complicated math problems and create a plot for a martial arts novel from scratch. The bot works mainly in Mandarin Chinese, its primary language. It is also able to handle queries and produce responses in English at a less advanced level.

Li said the demonstrations showed how the bot had been “significantly improved” in terms of its understanding of queries, generation of complex responses and memory capabilities.

I wasn’t able to confirm whether the company used its own AI chip, the supposed A100 equivalent, Kunlun II to train ERNIE 4.0 or if they went with Nvidia’s A100s. According to online reports, Baidu seems to be leaning heavily on Nvidia for training its models, and that FT article from August mentioned significant orders for A800 chips. As I mentioned in August, this will make for an intriguing case study in the future. If Chinese tech giants manage to develop a globally leading LLM, they'll have done it despite being somewhat limited by their tech choices. It's about to get pricier for them to compete, and it'll be even more remarkable if they pull it off (although I have my doubts).

Video Ads are hard(er)?

An article from The Information came out this week shedding light on some of the challenges Meta is facing in their efforts to monetize Reels.

Anyone scrolling quick videos of people making elaborate kinds of spaghetti or sharing dating horror stories may not be in the mood to click on an ad suggesting they buy a new pair of jeans. That’s at least what ad executives are saying about Meta Platforms’ efforts to sell ads on Reels, the Facebook owner’s answer to TikTok.

A year after Meta began selling ads on Reels, the company is struggling to convince marketers that advertising on the TikTok rival can drive new business, ad executives say. Marketers typically view Meta’s properties as venues to run ads designed to persuade people to do something, like buy a product or download an app, whereas quick video ads that run on Reels are better to promote a brand.

One executive at a major agency that spends hundreds of millions annually on Meta’s platforms said in an interview that his clients mostly don’t view Reels as a strong option for direct-response advertising, compared to placing ads on Instagram and Facebook’s feeds. Roughly 45% of the firm’s clients currently advertise on Facebook Reels, which has stayed mostly flat since the beginning of the year, the executive said, indicating it is not seen as a major priority. He said he was seeing more growth on Instagram Reels, where roughly 65% of the firm’s clients currently advertise, compared to roughly 55% at the beginning of the year.

At the same time, Meta appears to be slowing its efforts to promote Reels to advertisers. Ad sales representatives at Meta were aggressively pitching Reels to advertisers up until earlier this year, according to a Meta employee. However, that pitch has taken a backseat of late as Meta shifts its focus toward getting more advertisers to use its artificial intelligence tools that decide where to deliver ads.

“Everything they can talk about [now] is all AI, AI, AI,” said Bryan Karas, the CEO of Playbook Media, a marketing agency focused on helping brands advertise on Meta apps, who said he has noticed a sharp shift in messaging from this time last year. Karas also said the performance of Reels ads had improved in that time.

Inside Meta, the company has struggled to nail down the right business model for Reels, multiple current and former employees said. When Meta first launched ads on Reels, it tested “sticker ads” that covered the top of the videos, according to a former employee. But the ads were not popular with either brands or users, the person said. More recently, the company experimented with “overlay” ads that covered the bottom of Reels videos, later expanding that test to Instagram.

For now, the company appears largely set on running ads between individual videos—a recent scroll showed an ad encouraging people to download an app that tells users if their plants are healthy or not, as well as an ad featuring the founders of a chai company explaining how drinking bad chai in America inspired them to start their business. Meta is also encouraging advertisers to work with creators to make video ads such as clips of influencers trying on different outfits from clothing brands, a current employee said.

But relying on running ads between videos also has problems, mainly because the way people use Facebook and Instagram is quite erratic. Many people come to Reels videos by accident: a video might appear in their Instagram feed, for example, or a friend might send them a video over direct message. In those situations, the videos appear in isolation, meaning there is no option to play ads between a chain of videos.

Complicating matters, most top advertisers and ad agencies have separate teams focused on video, which prioritize TV and streaming services, and social platforms, which are earmarked for Meta’s apps and rivals such as TikTok and Snap. And the ad buyers on the video team typically don’t know enough about short-form video, including TikTok, to be willing to move dollars away from TV and streaming, said one ad executive.

It doesn’t help that Reels is also just the latest effort by Meta to break through in video advertising: Facebook Watch, which once hoped to rival YouTube, and IGTV were built to lure more video viewers and ad spending, only to fail at capturing either in any meaningful way.

The digital and social teams at the big advertisers, meanwhile, continue to prioritize platforms they’re familiar with that can be the most effective in driving performance.

When people are watching videos they are in a “lean-back” state, according to Karas, and have “less intent and less willingness to click off and actually go purchase something.” That’s bad news for Meta. “If you can't directly tie that ad impression to an outcome, advertisers will decide that [ad] has less value,” Karas said.

Meta’s artificial intelligence systems, which deliver ads to different areas on Meta’s apps based on predictions about what ads will compel users to buy products, may not be helping.

Many advertisers buy auto-placement package deals with Meta, so their ads are automatically placed across different parts of Instagram and Facebook, including news feeds, Instagram Stories and Reels. However, in some cases, the AI that powers Meta’s system has been sending only a small fraction of those ads to Reels, frustrating sales representatives, the employee said.

These are all fair points, though I would attribute part of the struggles due to short form video (SFV) being a nascent format for advertisers. I don’t necessarily agree with the notion that only brand ads can be run on SFV. So far that’s been the dominant ad shown on TikTok because the company hasn’t built as good as an ad targeting platform as Meta. Meta has the best ad targeting in the industry. Instagram and FB Blue have collected a ton of data from its users’ profiles (even basic things like sex and age that TikTok may not have) and continue to learn more about their preferences whenever they interact with performance based ads. This puts them in a better position to monetise SFV than competitors. In addition, the ads themselves aren’t that different from those shown on Stories (so it shouldn’t be that much extra work to implement). I see both display and video ads when scrolling through Stories or Reels. They may not monetise at the same rate because it takes longer to scroll through Reels (and hence there’s less time to serve ads), but the company says it’s driven incremental time spent on the app. Instagram users are only using the app about 60% of the time of TikTok, so there’s room for impressions to grow if consumption continues to shift towards SFV. It will take time though, Stories was introduced in 2018 and only reached monetisation parity with Feed about four years later. I can also understand why the messaging from ad sales reps has shifted away from Reels and towards AI. Pushing advertisers onto Advantage+ will ultimately drive greater adoption of Reels as the ad buying managers (who may be hesitant on SFV) take a backseat and let an algorithm allocate dollars to whatever format generates the best return on ad spend.

Meta isn’t the only company finding its feet selling a new ad format. I wrote the other week about how Amazon is looking to introduce ads to its standard tier on Prime Video in 2024. This article from The Information details some of the tactics Amazon is deploying to boost the adoption of its video ad placements.

Amazon is approaching brands that already buy search ads on its e-commerce website with an unusual pitch: if you buy streaming TV ads, we’ll make them for you.

The incentive highlights the lengths Amazon is willing to go to in order to try to build a streaming ad business. Smaller brands that sell on Amazon often don’t have the money or expertise to make video commercials, and that typically makes them a less attractive source of business for an ad seller. In response, Amazon in recent months has been paying for the writing, shooting and editing of commercials if brands spend at least $15,000 to run them on Amazon streaming sites like Freevee and Twitch, according to two marketing consultants whose clients have taken Amazon up on the offer.

Amazon may feel it has little choice but to try to woo small merchants. It is trying to build a streaming ad business at a time when nearly every major streaming service–including the ones owned by TV companies with long experience in advertising–has begun selling ads. The e-commerce giant has an edge in one area: the millions of merchants that sell on its e-commerce platform, many of which already buy search ads sold by the company that give their products a leg up in search results. Amazon appears to be hoping to replicate that search ad success in streaming. The challenge is, however, that streaming video ads are typically pricier than search ads.

One commercial that Amazon produced for a maker of safety devices for seniors shows actors gardening, walking and watching television while using the devices. Viewers can scan a QR code in the corner of the screen that directs them to the brand’s Amazon page, and the end of the 30-second spot features the Amazon logo underneath the brand’s logo.

However, there are signs that not every brand that’s taken up Amazon’s offer to produce commercials has been happy with the end product.

One household products brand that already spends about $250,000 a month on traditional television ads decided to take Amazon up on its offer to produce a commercial for its streaming sites in exchange for a spending commitment, said Jason Boyce, a consultant with Avenue7Media who advises the brand. The commercial Amazon produced was lower quality than the brand’s television ads, and the brand decided to stop paying to run it after it had reached the minimum it had agreed to spend.

“The content that was created didn’t really differentiate the brand—it was a little too generic,” Boyce said. “The production quality wasn't as good as they were hoping, but it was free.”

Still, Boyce said that the brand is considering buying more Amazon streaming ads in the future, likely using commercials it produces itself. And while Boyce said that he expects plenty of other brands to explore video ads through Amazon, he said that streaming ads are only likely to be worth it for brands that can spend tens of thousands of dollars per month for several consecutive months.

“You need to have repetition serving those TV ads because running for a week or two doesn't create mindshare,” Boyce said. “Without a significant budget to continually show and serve those ads to gain that mindshare, it’ll be really hard for the average small brand.”

Video ads tend to be costlier due to content creation expenses and the ad slot's duration compared to search and display ads. The return on ad spend is often lower because viewers are less likely to take action during movies or TV shows, except possibly during sports. This limits video ads primarily to larger businesses, drawing funds from the traditional media market instead of expanding the ad industry, as Google and Facebook have. While GenAI tools can reduce content and production costs, it's possible that these savings will be offset by higher CPMs in auctions. Nonetheless, even if smaller merchants drove a modest low single digit percentage boost to the $350 billion ad market, it would be a lucrative opportunity. Amazon, with its rich customer data, is well-positioned to gain share from traditional media.

Related: Instagram took away a major source of income for its creators, and they're not happy about it

That’s all for this week. If you’ve made it this far, thanks for reading. If you’ve enjoyed this newsletter, consider subscribing or sharing with a friend

I welcome any thoughts or feedback, feel free to shoot me an email at portseacapital@gmail.com. None of this is investment advice, do your own due diligence.

Tickers: MSFT 0.00%↑ GOOG 0.00%↑ , NVDA 0.00%↑, BIDU 0.00%↑, META 0.00%↑, TikTok, AMZN 0.00%↑