AI frenzy sweeps sovereign capital! Indonesian sovereign wealth fund joins $3 trillion AI infrastructure race.

Indonesia's first sovereign wealth fund is currently investing in a series of funds and private companies, with a focus on investing in AI infrastructure for computing power. They are following global sovereign institutions like PIF in Saudi Arabia in embracing the unprecedented AI investment frenzy. Indonesia's sovereign wealth fund is accelerating its investment in AI data center infrastructure, highlighting the AI investment frenzy sweeping across Southeast Asia, as well as the AI super bull market expanding from "large tech giants" to various layers of assets related to global AI data center construction. It is understood that the Indonesia Investment Authority, along with its co-investors, have deployed around 74.5 trillion Indonesian rupiah (approximately 4.20 billion USD), with about 30% of this going towards digital infrastructure assets closely related to AI. Christopher Ganis, the Chief Investment Officer of the Indonesian sovereign wealth fund, mentioned in an interview. Ganis stated that Indonesias INA, with assets under management exceeding 8 billion USD, has completed several digital infrastructure transactions, including participating in a financing round for DayOne Data Centres Ltd. It is known that DayOne Data Centres Ltd. is a company spun off from GDS Holdings Ltd., a leading data center company in China. Indonesia's sovereign wealth fund enters the AI data center arena This sovereign wealth fund's significant investment in AI computing infrastructure highlights the global rush to invest in super-sized AI data centers and other beneficiaries of the machine learning revolution. According to Moodys Ratings, at least 3 trillion USD will flow into data center-related investments in the next five years, offering tremendous investment opportunities for funds like INA. In the eyes of seasoned market analysts from Wall Street banks like Morgan Stanley and Nomura, the main theme of the global AI super bull market, including US stocks, is shifting from a "NVIDIA AI GPU demand surge" to a "bullish wave across the AI computing infrastructure chain". This will continue to support the super bull market surrounding AI in the stock markets worldwide. A prime example is NVIDIA's recent performance, which clearly demonstrates that the global AI computing infrastructure construction frenzy is far from stopping, extending from AI GPU/AI ASICs to data center CPUs, high-performance network infrastructure, enterprise-level HBM/DRAM/NAND storage, machine-level server clusters, AI super factories, and enterprise-level large-scale AI cloud computing systems. Analysts on Wall Street are increasingly bullish on NVIDIA, with an average target price suggesting NVIDIAs market value could exceed 7 trillion USD, and industry-wide AI training/inference computing infrastructure expenditure is expected to reach at least 3 trillion USD by 2030. Ganis stated that the digital wave associated with AI is one of the five key focus areas for this five-year-old fund, with an annual maximum of 30% of the allocation going into this sector. He mentioned, "Even though many trends related to AI come from outside Indonesia, it does not mean we will skip this trend." INA expects Indonesia to benefit from this AI wave as some tech companies bring large AI training and research centers to Indonesia. Indonesia has one of the youngest populations in Asia. According to its website, the fund's portfolio includes several data centers planning a capacity of 74 MW with a promised occupancy rate of 100%. Amidst intense competition between the US and China in the field of AI, Ganis mentioned that INA will adopt a "non-aligned" approach in collaborating with diversified partners. Ganis stated that AI is an important industry in a "multi-polar world", and countries will strive to create national or regional champions in the field. However, he also added that businesses might see Indonesia as a market for improving productivity and expanding profits, rather than solely a competitive battleground. Despite geopolitical conflicts in the Middle East putting pressure on Indonesia, a net importer of oil, Ganis mentioned that the country's existing data centers are located in core industrial zones with a secure power supply system. However, he remains cautious about the potential further rise in oil prices. Ganis remarked, "I don't think you can make a blanket statement for all the impacts," adding that the fund must assess various factors, including future energy contracts, to determine their impact. Established in December 2020, this sovereign wealth fund recently underwent a series of leadership changes. Its peer, Danantara, has much broader authorization scope and is also seeking investment in AI data centers. Initially focusing on minority equity investments in private equity, INA has since expanded into private credit and real estate sectors. The sovereign investment tool was initially established with government equity holdings and cash injections, relying heavily on high dividend stock payouts and interest income. Ganis anticipates that as investment income proportion increases, this part of the portfolio will gradually decrease. AI super bull market enters the "full hardware chain reassessment" phase When a major Southeast Asian country like Indonesia allocates up to 30% of its annual allocation to digital infrastructure and bets on data centers to massively accommodate AI training, inference, and research center demands in the future, this reinforces the AI super bull market's expansion from "large tech giants" to various layers of assets related to global AI data center construction, including storage chips, high-performance network equipment, power equipment, liquid cooling systems, optical communication/optical interconnects, advanced packaging, server ODMs, and assets related to REITs data center assets/private credit assets. As AI intelligence sweeps the globe, the main theme of AI computing investment is shifting from a "competition surrounding AI GPU" to an "AI intelligence-driven full-stack computing system". The next round of excess alpha returns will no longer only belong to the strongest leaders in the GPU/ASIC domains, but will systematically diffuse into data center CPUs, DRAM/NAND/HBM storage, AI PCBs, liquid cooling systems, data center optical interconnect systems, ABF carriers/glass substrates, and extensive wafer foundries in the full-stack AI computing infrastructure layer. Thus, the "cake" of the AI super bull market is expanding from GPU core assets to the full-stack AI hardware bottleneck chains. NVIDIA still sets the tone for the AI bull market, but PCBs, MLCCs, ABF, ODMs, liquid cooling, power supplies, silicon wafers, CMP consumables, photolithography materials, glass substrates, SOI/InP, data center optical interconnect equipment, optical modules, and advanced packaging equipment, could all undergo a revaluation in this new cycle. While the previous market phase mainly saw the purchase of GPUs, HBM, cloud capital expenditures, and resilience in NVIDIA's performance, a recent research report from Morgan Stanley shows that the value of next-generation AI racks is soaring, not driven solely by GPUs, but also by synchronous increases in the values of PCBs, MLCCs, ABF substrates, power supplies, liquid cooling, ODM assembly testing, etc. Morgan Stanley estimates that the Rubin VR200 NVL72 rack price is around 7.8 million USD, nearly double the GB300 price of around 3.99 million USD; with the PCB content value increasing by 233%, MLCC by 182%, and ABF substrate by 82%. This indicates that the AI super bull market is entering the second phase of "component content expansion". From an engineering perspective, Rubin is not just about GPU replacement, but a leap in rack-level system complexity. When NVLink/NVSwitch, ConnectX, BlueField DPU, midboards, switch trays, SOCAMM/high-bandwidth memory, and liquid-cooled power supplies all come together in a single rack, the value chain naturally diffuses from GPU chips to high-speed copper connections, data center optical interconnect systems, power supply, cooling, substrates, and high-layer PCBs. The proportion of GPUs in BOM actually decreases from about 65% in GB200 to about 51% in VR200, but the absolute amount for GPUs still rises from around 2.52 million USD to about 3.96 million USD; this is the most crucial change in the AI industry chain: the leaders remain strong, but the "supporting links" are being revalued as indispensable performance bottlenecks.