Google Expands AI Chip Line, Pressure on Nvidia Mounts
Baca dalam 60 detik
- Google akan meluncurkan dua unit tensor processing (TPU) khusus tahun depan, menargetkan pasar komputasi rahasia dan aplikasi data besar.
- Bloomberg Intelligence memperkirakan penjualan TPU Google bisa melampaui US$135 miliar pada 2028, dengan harga 30% lebih murah dari GPU Nvidia.
- Nvidia kehilangan lebih dari US$250 miliar nilai pasar setelah dewan mengumumkan investigasi atas dugaan CEO Jensen Huang menjadi target intelijen asing.

Google has officially expanded its artificial intelligence (AI) chip lineup with plans to introduce two dedicated tensor processing units (TPUs) next year. The move intensifies competition in the AI accelerator market long dominated by Nvidia, just as the chip giant faces internal turmoil over an investigation into CEO Jensen Huang.
The two new TPUs are designed for different segments. The first variant targets confidential computing for customers handling sensitive information, while the second supports high-volume data applications such as software development and drug discovery. According to Bloomberg Intelligence analysts, Google's TPU sales—including supporting hardware and software—could exceed US$135 billion by 2028, up sharply from an estimated US$12 billion this year. That figure underscores Google's ambition to be more than a supporting player in the AI ecosystem.
What makes the pressure on Nvidia more palpable is the price factor. Bloomberg Intelligence estimates Google's TPUs are about 30% cheaper than Nvidia's GPUs. That cost gap could be a major draw for companies looking to develop AI models without inflating capital expenditure. If Google can convince customers that TPU performance is on par, Nvidia's dominant market share could erode.
On the other side, Nvidia has been hit by negative sentiment from within. The board of directors announced an investigation following reports that Jensen Huang was a target of foreign intelligence. Huang acknowledged having an affair with a woman whom US officials suspect was recruited by Chinese intelligence. The investigation will be overseen by independent directors and outside counsel. Nvidia said it found no evidence of company secrets being leaked or of a threat to national security, and the board said it continues to support Huang, who has promised to cooperate.
Nevertheless, the market reacted sharply. Nvidia's market capitalization fell by more than US$250 billion in Monday (9/11) trading, and its shares plunged before paring losses to 4.3%. The anomaly was that the Nasdaq index actually gained more than 2%, including a 3.4% rise in Google shares. The phenomenon shows that market participants are beginning to distinguish between issuer-specific risks and the outlook for the AI sector as a whole.
The pressure on Nvidia is not only about chip prices. In the broader market, neocloud and AI infrastructure stocks such as IREN, Cipher Digital, and CoreWeave have collapsed 70% or more from their peaks. This signals that euphoria over the AI trade is starting to correct, and investors are shifting toward players with stronger fundamentals and more efficient operating costs. Google, with its large cash pile and established cloud ecosystem, is well positioned to capitalize on this shift.
For Indonesia, the dynamics of AI chip competition are more than a spectator sport. Plans to build data centers and national AI sovereignty depend heavily on the availability of affordable accelerators. If Google's TPUs are indeed 30% cheaper, domestic cloud service providers could lower infrastructure costs and accelerate AI adoption in both the public and private sectors. However, dependence on one or two global suppliers remains a strategic risk that needs to be anticipated through multi-party partnerships.
Going forward, the battle between Google and Nvidia will be tested next year when the new TPUs begin to be marketed. The question is whether Nvidia can maintain its price premium and market share amid a wave of more economical products, or will be forced to adjust its pricing strategy to retain customers. What is clear is that the era of single dominance in the AI chip market appears to be ending.



