Amid this whirlwind of upheaval and excitement, however, nothing has dominated recent market discourse quite like the sharp shift in sentiment away from the software sector — an episode that is currently still unfolding, and which has been widely dubbed the “SaaSpocalypse”.
The SaaSpocalypse
At the end of January, AI startup Anthropic (the developer of the Claude line of AI models that competes with OpenAI’s ChatGPT) released a series of plugins for its newly launched general-purpose AI agent, Cowork. The plugins provide ready-to-use commands designed for specific workflows, including legal research and financial analysis, sparking debate over how rapidly AI is beginning to automate highly specialised professional tasks.
The legal plugin, in particular, appeared to heighten market anxiety, sending the legal-tech sector into a heavy sell-down. The week that the plugins were unveiled saw bellwether companies within the sector, such as Thomson Reuters and RELX (the parent company of legal database LexisNexis), falling roughly 20%; the latter experiencing its steepest single day drop since 1988. This was soon followed by a sell-off that rippled across the IT services sector and broader SaaS (software-as-a-service) market.
See also: Investing takes intelligence, divesting takes wisdom
It would be misleading, however, to attribute souring software sentiment solely to the launch of the Claude plugins. While their launch may have triggered the sell-off, they are far from the sole cause. The truth is that AI displacement fears have weighed on the sector for some time, and market performance of software companies has lagged the broader S&P 500 index over the past year (see Chart 1).
AI anxiety
Central to this shift in software sentiment is the growing concern that advances in AI could render long-established SaaS offerings obsolete. The recent rise of “vibe-coding” — the practice of using AI tools such as Cursor, Claude or ChatGPT to write code and build software — has only amplified these fears. After all, if software can be built on demand and at far lower costs, why pay the recurring premium for an off-the-shelf SaaS product?
See also: The false dichotomy: Allowing yourself no option because of fear brings the worse outcome
There are persuasive counterarguments, of course. Contrarians point out that established SaaS offerings are more equipped to handle complex cybersecurity and compliance requirements; and that, as the term SaaS implies, these companies provide more than software alone, but also deliver ongoing services — maintenance, updates, operational support and so on. In other words, there appear to be a host of reasons for why ad-hoc AI-generated code from the new wave of AI tools cannot be expected to replicate the full value that SaaS companies provide over the long term. Jensen Huang, for one, has called the idea that software is in decline “the most illogical thing in the world”.
In any case, for us the question of whether Claude Cowork can truly function as a digital coworker is secondary. There is currently no way to answer this with any certainty. But even without a definitive answer, what matters is that the risk-reward profile for companies in the software sector has shifted materially, and that shift carries clear implications for our AI portfolio. On this basis, we have made several adjustments to the portfolio composition.
AI Portfolio — SELLS
It requires no crystal ball to foresee that AI will have a far-reaching and profound impact on the global economy in the years ahead. However, what is far less certain is the path by which that future unfolds. As we have noted in the past, the AI portfolio is exploratory in nature, evolving as we draw lessons from the rapidly changing tech landscape.
The first of these lessons is still actively unfolding through the ongoing SaaSpocalypse. In the current environment, where the market has latched onto a powerful narrative (“software is dead”), pushing against the tide is rarely a winning strategy. As such, we have sold ServiceNow and Twilio, our primary SaaS exposures.
Separately, we have also exited Horizon Robotics, a Chinese ADAS chip manufacturer that primarily serves the domestic EV market. While China’s domestic substitution push should in principle benefit local chip suppliers such as Horizon Robotics, other market developments are less positive. China’s EV sector is expected to underperform in 2026, weighed down by flagging demand post-stimulus withdrawals and oversupply, as well as rising production costs. In addition, BYD, a key customer, plans to adopt home-grown chips by mid-2026, posing a significant revenue risk for Horizon Robotics.
AI Portfolio — HOLDS
For more stories about where money flows, click here for Capital Section
Worth noting, while we have reduced our broader software exposure, we continue to hold two positions within the sector: Cadence Design and Datadog. Unlike the broader software market, these companies occupy positions that enable AI adoption rather than being threatened by it. Electronic design automation (EDA) company Cadence Design benefits from increasing chip design complexity and the ongoing data centre buildout, whereas Datadog is positioned within an industry (observability and monitoring) with strong structural growth catalysts. Both companies are increasingly important players in the AI era.
Beyond these two positions, the remainder of the AI portfolio sits within the infrastructure and hardware sectors. In other words, the “picks and shovels” of the AI era. Our US exposure is anchored by Amazon and Marvell Technology, providers of essential data centre technologies that power AI growth. Our China exposure is represented by Alibaba, Naura Technology and RoboSense: Alibaba, at the cloud and compute layer; Naura Technology, at the semiconductor manufacturing layer as a key chip manufacturing equipment supplier; and RoboSense, as an industry leading LiDAR and sensing solutions provider for the EV and embodied AI end-markets.
AI Portfolio — BUYS
Consistent with the theme of our existing holdings (hardware and infrastructure), we have redeployed capital from the recent portfolio sales into three high-conviction names, each offering exposure to segments that we believe combine resilience with upside in current market conditions: Minth provides exposure to the EV supply chain; Sieyuan Electric, to global electrification and the power grid; and Unusual Machines, to the US defence and drone industry.
Minth
Despite near-term weakness in China’s EV market, the sector remains a key driver of the nation’s broader industrial innovation. As we have mentioned before, China’s approach to innovation often arises from using existing industrial supply chains to support the commercialisation of new breakthroughs. In the case of EVs, the industry was historically built upon China’s strengths in consumer-grade lithium-ion batteries and large-scale auto manufacturing.
Today, a similar pattern is unfolding: leading automakers have begun branching into emerging fields such as humanoid robotics and eVTOL (electric vertical takeoff and landing) aircraft, while their upstream suppliers have begun adapting their manufacturing capabilities as well. As such, we believe it prudent to retain exposure to this sector. In replacing Horizon Robotics, we focus on upstream suppliers with strong growth potential, of which Minth stood out as a top pick.
Currently, the bulk of Minth’s revenue comes from its traditional automotive business, which is focused on structural exterior parts (plastics, aluminium and trim components) and battery housing. The company stands out among peers for its strong European footprint (see Chart 2), holding over 30% market share of EU’s battery housing market. This places Minth in a favourable position to benefit from accelerating EV adoption in Europe, a region that currently offers stronger growth prospects than either China or the US.
Beyond the auto end-market, Minth has also begun expanding into emerging high-growth markets: humanoid robotics, eVTOL aircraft and AI server cooling. Building on its longstanding manufacturing expertise, the company has partnered with leading players such as AgiBot (humanoid robotics) and Ehang (urban air mobility) to accelerate product development and capture early market share. As for liquid cooling solutions for data centres, deliveries to Taiwanese AI server customers commenced at the end of 2025.
In 1H2025, Minth reported revenue and net income growth of 11% and 20% year on year (y-o-y) respectively. Revenue from international markets grew 22%, driven by growing demand for structural parts and battery housings from EU.
Sieyuan Electric
Sieyuan Electric is a Chinese power equipment manufacturer and engineering services provider that produces power grid equipment for both the Chinese and international markets. As such, the company is well-positioned to capture growth on multiple fronts.
Domestically, Sieyuan Electric is a key beneficiary of China’s ongoing grid buildout. The State Grid, China’s largest grid operator and a major customer of Sieyuan Electric, recently announced plans to invest RMB4 trillion under the 15th Five-Year Plan (2026–2030), representing a 40% increase from the previous investment cycle. Beyond traditional grid infrastructure, the company has also partnered with leading lithium battery manufacturer CATL to develop energy storage systems (ESS) and supporting technologies — solutions that are increasingly critical as the rising share of renewables in China’s grid introduces greater volatility in power supply, making ESS essential for maintaining the overall stability of the grid.
In international markets, Sieyuan Electric is similarly well-positioned, benefiting from the ongoing electrification of emerging economies in Africa and Southeast Asia, as well as the rapidly rising energy demands in the AI era, which call for more robust and better-equipped power grids. The company’s overseas revenue grew at a 27% compound annual growth rate from 2020 to 2024 and has since accelerated significantly.
In the first nine months of 2025, Sieyuan Electric reported strong results, with revenue and net income rising 33% and 47% y-o-y respectively, mainly attributable to accelerating international growth. In December 2025, it was reported to be exploring a dual listing on the HKEX, with a tentative timeline set for the second half of 2026.
Unusual Machines
Unusual Machines is a small-cap US company that manufactures drones and drone components. The company’s consumer-facing business operates through the Fat Shark and Rotor Riot brands, serving FPV (first person view) and hobbyist markets. Separately, the company has a B2B channel selling NDAA (National Defense Authorisation Act)-compliant parts directly to original equipment manufacturer (OEM) suppliers of the US military. It is this latter segment that has emerged as the company’s primary growth driver.
In recent months, the company has made significant strides towards establishing itself as a vertically integrated drone parts supplier for the US military. Beyond expanding its existing facilities and headcount, Unusual Machines has engaged in a number of strategic investments, including two ongoing reverse mergers: software company XTEND with JFB Construction Holdings, as well as drone manufacturer Powerus with Aureus Greenway Holdings. The XTEND-JFB merger is expected to reach a market value of approximately US$1.5 billion upon listing, while the Powerus-Aureus Greenway deal, which was announced March 9, remains under development. XTEND is an Israeli software company that provides AI operating systems for drones whereas Powerus is a US drone manufacturer that develops autonomous drones for military and high-risk environments.
Unusual Machines is a confirmed supplier of two major federal UAS programmes: the Purpose Built Attritable Systems (PBAS) program for designing low-cost tactical drones for the US Army as well as the US$1.1 billion Drone Dominance Program, a specific procurement initiative funded by the One Big Beautiful Bill (OBBB). The company also stands to benefit from growing demand for NDAA-compliant drone parts through the Army-led SkyFoundry programme, which aims to mass-produce up to 10,000 small drones per month by the end of 2026.
While Unusual Machines remains loss-making on an annualised basis, it did achieve its first profitable quarter in 3Q2025, but this was mainly due to unrealised gains on short-term investments linked to its strong cash balance. For the full year, the company achieved y-o-y revenue growth of 101%, beating market expectations by a far margin. Earnings, however, came in under-expectations due to higher operating expenses driven by ongoing scaling and non-cash stock compensation. Nevertheless, it is projected to achieve full operational breakeven by late 2026 to early 2027, in line with the expected ramp-up of deliveries for enterprise orders.
Concluding remarks
While AI transformation is certain, winners are not. In such an environment, it is more likely that the risk-reward will favour adaptation. Strong conviction can be dangerous. It is clear to us that every time we think we know the AI winners, we are already late.
Portfolio commentary
The Malaysian portfolio traded marginally lower for the week ended March 25. Hong Leong Industries (+1.3%) and Kim Loong Resources (+1.2%) gained while Maybank (-1.4%), United Plantations (-1.2%) and LPI Capital (-0.1%) were the losing stocks. Total portfolio returns now stand at 215.1% since inception. This portfolio is outperforming the benchmark FBM KLCI, which is down 6.2% over the same period, by a long, long way.
The Absolute Returns Portfolio fell 2.8% in the same week, paring total portfolio returns to 34.4% since inception. The top losers were SPDR Gold Minishares Trust (-6.4%), Alibaba (-6.2%) and Ping An - A (-5.2%). We continue to hold a relatively high level of cash in view of the prevailing uncertainties.
The AI portfolio also ended in the red last week, down 4.8%. Total portfolio loss now stands at 5.7% since inception. The three gaining stocks were Marvell (+12.4%), Amazon (+0.9%) and Naura Technology (+0.1%), while the biggest losers were Unusual Machines (-26.1%), RoboSense (-6.5%) and Alibaba (-6.2%).
Disclaimer: This article is strictly for information purposes only and does not constitute a recommendation or solicitation or expression of views to influence readers to buy or sell stocks, including the stocks mentioned. Any personal investments should be made at the investor’s own discretion and/or after consulting licensed investment professionals, at their own risk. The author of this article does not hold or own any stock(s) featured in this article or have a vested interest in it at the time of writing.
