Tech Thoughts Newsletter – 12 January 2024.
Market: A typically volatile start to the year and a bit of a mean reversal to start (around concerns over rates and macro – new year, same problems!), though Nasdaq (and the fund) is now up year-to-date. Several layoffs announced in the tech sector (Google and Amazon most notably) are a reminder that macro momentum remains uncertain for 2024.
Portfolio: We made no major changes to the portfolio this week.
CES – the AI show
- CES is the largest tech show in the world, held in Las Vegas. While it’s still named “Consumer Electronics Show”, it’s fair to say that it has morphed fairly significantly (and with less buzz) over the past several years. This year, though, the buzz was back. Because, well, – AI.
- The most talked about product was from a company you’ve never heard of – the “rabbit r1”. It cost $200 and was pitched as a personal assistant device, much simpler than a phone and used for utility, not entertainment. Instead of finding and booking a restaurant using a series of apps on your phone, you’ll give your r1 a command in natural language.
- While r1’s compute happens in the cloud – via Nvidia GPUs (which makes it a tricky business – how quickly will compute costs outweigh the initial $200 revenue? The broader theme for CES was around AI compute on device, and in the context of the weaker PC numbers this month, the optimistic view remains that AI on devices will be a driver of the PC/smartphone replacement cycle this year.
- AMD’s keynote was “The future of AI in personal computers”, during which it announced its Ryzen 8000-series desktop processor, built to better handle AI. It comes with an integrated NPU (neural processing unit), which will effectively allow for more AI applications to take place on the end device. Intel also announced an integrated NPU last month at its “AI everywhere” event.
- For Nvidia, the focus was on its Hybrid AI platform, which will use Nvidia GPUs to use LLs and then PC GPUs with specialised AI Tensor cores to run low latency and cost-sensitive workloads.
- We’ve spoken before about AI compute at the edge – shifting AI workloads to the edge (either at the network edge or end device) makes sense, both from a cost and a latency perspective.
- Separately, while we’re on AI hype, India announced a $1bn investment in AI datacentres overnight, spending $500m on Nvidia GPUs.
Portfolio view: AI driving a new product cycle in PC remains a potential upside for AMD (owned).
December EV numbers – keep looking to China
- Tesla reported Q4 deliveries ahead of consensus, which meant for the full year Tesla achieved its 1.8mn volume goal for 2023.
- We’ve said a lot last year – and it will continue to be true this year – that the market misses a lot by focusing too much on the global auto OEMs and not enough on Chinese OEMs.
- Chinese December EV numbers came in ahead of consensus – December domestic retail sales tracking at 879k units (+14% mth/mth) and December EV wholesale volume (both export and domestic reaching 1.13m units (+18% mth/mth) – both ahead of consensus.
- BYD sold 341k vehicles in December, bringing 2023 sales to 3m (PHEV and EV) – now selling more battery-only vehicles than Tesla.
- While we talk a lot about BYD as the biggest EV producer, it’s worth noting that it’s no longer just the low-end/mass market that China’s EV is carving out. Brands like Li Auto and Zeekr (Geely-owned) continued their momentum – both at the higher end. Huawei’s luxury Aito M9 was reported to have reached >20k units in a week in December – quite an extraordinary number.
- It’s a further headache for Global OEMs selling at the higher end into the Chinese market (China is over a third of BMW’s total deliveries). Not only is there a share shift from international OEMs to Chinese, but also an accelerated destruction of the combustion engine market as a whole. The EV price war continues – Tesla is cutting pricing again in China this week.
- Also, this week, China has a new official target of 45% of new vehicles being NEV by 2027 (though we think the penetration level is pretty close to this already).
- The December US SAAR came in at 15.8m units – slightly ahead of consensus, with EV doing marginally better. As a reminder, the industry isn’t expected to show significant growth in 2024, with our thesis around auto semis much more around the shift to EV, and the increased semis content.
Portfolio view: There has been a generally negative sentiment in the market around EV demand, which these numbers help mitigate. The EV shift and the increased semiconductor requirements in EV continue to be a key exposure in the portfolio for us – we own Infineon and NXP, both of whom sell into BYD.
Mobileye warning – global OEMs losing share
- Mobileye (not owned) issued the first significant profit warning of the year in the technology sector. It lowered its guidance significantly due to an inventory correction.
- We suspect it is much more a market share issue for Mobileye than a broader issue in the auto market – with OEM adoption of its SuperVision solution undershooting expectations (and its global OEM customers themselves losing share – it doesn’t supply Tesla – and we believe BYD is mainly working with Nvidia. This week, Nvidia also announced partnerships with Li Auto, Zeekr and Xiaomi).
- Another warning – Microchip (not owned) said it expects the December quarter to come in at a 22% qtr/qtr decline – worse than the original expectations of down 15-20%, albeit not dramatically worse. Microchip is very exposed to industrial, and China, which is still the area of real weakness in the market. It is struggling to keep inventory levels low as customers continue to request to push out their backlog while trying to rebalance inventory themselves.
- We note that Microchip has some auto exposure, but we would point out that vs Infineon and NXP, Microchip doesn’t obviously benefit from the ICE-> EV shift in terms of semis content, and its components are much more dependent on headline auto volumes. The positives: management will lower utilisation further if they need to, and pricing hasn’t started to deteriorate yet.
Portfolio view: We do not think that the Mobileye miss is indicative of a broader industry problem. Infineon and NXP (which we own) have told us they are carefully watching their channel inventory – much more closely than ever before and have much more visibility on their customers’ inventory than in any prior cycle.
Memory momentum continues
- SIA announced November monthly semiconductor sales, above seasonality (again) – up 3.3% mth/mth vs usual seasonality 0.9%, driven by both units and pricing, in particular, analog and DRAM.
- After the worst memory downcycle since the financial crisis, DRAM pricing has started its upturn after production cuts earlier in the year began to feed through to better utilisation, compounded by the beginning of the ramp of HBM alongside AI demand.
- Relatedly, Samsung is rumoured to be raising its DRAM utilisation rate to over 80% from 60-70%, given the quicker-than-expected demand recovery.
- We’ve noted before the increased die size of moving to HBM, which limits the DRAM production growth (and is a positive for the semicap industry).
- Samsung’s Q4 results were weak, and the stronger memory ASP couldn’t offset the weaker foundry results. Samsung’s foundry business remains a struggle. The risk for us is that it moves to a competitive pricing model, but we think it will have difficulty gaining traction with clients.
Portfolio view: We don’t own memory players (they don’t meet our high return on invested capital requirements), but we will benefit from increased pricing and die sizes through our semicap exposure (in particular LAM, which is overexposed to memory).
Semicap – early Q4 results coming in positive
- VAT and Disco announced preliminary Q4 numbers that were better than guidance/expectations. VAT’s Q4 orders were up 44% qtr/qtr – confirming the bottoming of the semiconductor investment cycle.
- ASML announced that the US had asked to stop shipping its high-end DUV tool to China before the January deadline. This supposedly affected three tools, though will have no material impact on 2023 financials.
Portfolio view: Semicap equipment (we own Applied Materials, LAM Research, ASML and KLA) remains a key exposure in the portfolio for us. For 2024, we expect the market to grow again (single digit, second half driven), with leading-edge logic helped by the build-out of AI capacity; China continuing to spend at the trailing edge; and – perhaps the most critical swing factor – memory spend returning after the worst downturn in a decade (helped by AI-driven HBM demand).
Monthly numbers – TSMC, Hon Hai and Apple in line, notebooks weaker
- TSMC’s December revenue came in at TWD176bn, down 14% mth/mth and 8% yr/yr – though worth remembering December’s typical “down” seasonality. This means that Q4 revenue comes in ahead of consensus. FY23 revenue was down 4.5%, in line with the guide given back in April.
- Broadly, Apple suppliers (of which TSMC is one) have reported okay December quarters, with an “okay to mixed” Q1 outlook. Hon Hai reported its December sales number, bringing Q4 ahead of expectations for all its products (Apple’s iPhone being the biggest exposure). It guided Q1 to align with usual seasonality – it means no big surprises for Apple.
- Largan (another Apple supplier) also beat Q4 revenue, with cameras remaining a key feature that smartphone OEMs (including Apple) continue to upgrade.
- At the trailing edge foundry UMC December revenue dropped a bit from the NTD19bn levels we’ve seen – a per-usual seasonality. It seems that trailing-edge pricing is deteriorating.
- December notebook shipments came in -6% mth/mth and -3% yr/yr, which brought Q4 shipments to -3% yr/yr (and total 2023 shipments -13% yr/yr). The inventory correction might be behind us, but it’s clear that end-demand continues to be subdued. Q1 will be held back by seasonality and fewer working days, possibly compounded by the weaker macro – we expect a continued subdued environment overall (notwithstanding – as above – an AI-driven replacement cycle).
Portfolio view: No big surprises in the monthly data. For TSMC, we believe AI strength in the back half of the year has made up for continued damp momentum on smartphone and a longer-than-expected inventory correction. Going forward, we expect more capacity to move to AI and for that to drive sustained revenue momentum.
IT budgets/Indian IT services – more of the same
- Indian IT services companies Tata and Infosys are among the first to report (okay results). More important for us is the broader commentary around budgets and spend. Generally it feels like more of the same – continued budget scrutiny but not getting worse.
- TCS said: “Given the overall uncertainty, we find our clients also very agile, like even though they may have some thought in terms of what they want to spend, we find that they keep also reacting to the market sentiment.”
- Infosys commented: “But we don’t see any change in what we were seeing in terms of behaviour from the last quarter where budgets would suddenly have a different direction. So at this stage, it looks like it’s similar to what we were seeing, but everything is not yet closed out from our discussions on the budgets.”
Portfolio view: We don’t own any of the IT services players (which we view as being late-cycle and at risk of rolling over, and there are now many uncertainties around how AI will impact time and materials business models).
For enquiries, please contact:
Inge Heydorn, Partner, at inge.heydorn@gpbullhound.com
Jenny Hardy, Portfolio Manager, at jenny.hardy@gpbullhound.com
Nejla-Selma Salkovic, Analyst, at nejla-selma.salkovic@gpbullhound.com
About GP Bullhound
GP Bullhound is a leading technology advisory and investment firm, providing transaction advice and capital to the world’s best entrepreneurs and founders. Founded in 1999 in London and Menlo Park, the firm today has 14 offices spanning Europe, the US and Asia.